From internet-drafts@ietf.org Wed Sep 25 22:15:12 2013 Return-Path: X-Original-To: dmm@ietfa.amsl.com Delivered-To: dmm@ietfa.amsl.com Received: from localhost (localhost [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id E34A011E814C; Wed, 25 Sep 2013 22:15:11 -0700 (PDT) X-Virus-Scanned: amavisd-new at amsl.com X-Spam-Flag: NO X-Spam-Score: -102.561 X-Spam-Level: X-Spam-Status: No, score=-102.561 tagged_above=-999 required=5 tests=[AWL=0.039, BAYES_00=-2.599, NO_RELAYS=-0.001, USER_IN_WHITELIST=-100] Received: from mail.ietf.org ([12.22.58.30]) by localhost (ietfa.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id zTNDHaTZgQ2Z; Wed, 25 Sep 2013 22:15:11 -0700 (PDT) Received: from ietfa.amsl.com (localhost [IPv6:::1]) by ietfa.amsl.com (Postfix) with ESMTP id 21E0A11E8140; Wed, 25 Sep 2013 22:15:11 -0700 (PDT) MIME-Version: 1.0 Content-Type: text/plain; charset="utf-8" Content-Transfer-Encoding: quoted-printable From: internet-drafts@ietf.org To: i-d-announce@ietf.org X-Test-IDTracker: no X-IETF-IDTracker: 4.72 Auto-Submitted: auto-generated Precedence: bulk Message-ID: <20130926051511.30938.32880.idtracker@ietfa.amsl.com> Date: Wed, 25 Sep 2013 22:15:11 -0700 Cc: dmm@ietf.org Subject: [DMM] I-D Action: draft-ietf-dmm-requirements-08.txt X-BeenThere: dmm@ietf.org X-Mailman-Version: 2.1.12 List-Id: Distributed Mobility Management Working Group List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Thu, 26 Sep 2013 05:15:12 -0000 A New Internet-Draft is available from the on-line Internet-Drafts director= ies. This draft is a work item of the Distributed Mobility Management Working G= roup of the IETF. Title : Requirements for Distributed Mobility Management Author(s) : H Anthony Chan Dapeng Liu Pierrick Seite Hidetoshi Yokota Jouni Korhonen Filename : draft-ietf-dmm-requirements-08.txt Pages : 20 Date : 2013-09-25 Abstract: This document defines the requirements for Distributed Mobility Management (DMM). The hierarchical structure in traditional wireless networks has led primarily to centralized deployment models. As some wireless networks are evolving away from the hierarchical structure, such as in moving the content delivery servers closer to the users, a distributed model for mobility management can be useful to them. The IETF datatracker status page for this draft is: https://datatracker.ietf.org/doc/draft-ietf-dmm-requirements There's also a htmlized version available at: http://tools.ietf.org/html/draft-ietf-dmm-requirements-08 A diff from the previous version is available at: http://www.ietf.org/rfcdiff?url2=3Ddraft-ietf-dmm-requirements-08 Please note that it may take a couple of minutes from the time of submission until the htmlized version and diff are available at tools.ietf.org. Internet-Drafts are also available by anonymous FTP at: ftp://ftp.ietf.org/internet-drafts/ From h.anthony.chan@huawei.com Wed Sep 25 22:35:02 2013 Return-Path: X-Original-To: dmm@ietfa.amsl.com Delivered-To: dmm@ietfa.amsl.com Received: from localhost (localhost [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id 3A14921F9343 for ; Wed, 25 Sep 2013 22:35:02 -0700 (PDT) X-Virus-Scanned: amavisd-new at amsl.com X-Spam-Flag: NO X-Spam-Score: 0 X-Spam-Level: X-Spam-Status: No, score=x tagged_above=-999 required=5 tests=[] Received: from mail.ietf.org ([12.22.58.30]) by localhost (ietfa.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id CpqmZxLfxp79 for ; Wed, 25 Sep 2013 22:35:01 -0700 (PDT) Received: from lhrrgout.huawei.com (lhrrgout.huawei.com [194.213.3.17]) by ietfa.amsl.com (Postfix) with ESMTP id EAE9E11E8136 for ; Wed, 25 Sep 2013 22:34:57 -0700 (PDT) Received: from 172.18.7.190 (EHLO lhreml203-edg.china.huawei.com) ([172.18.7.190]) by lhrrg02-dlp.huawei.com (MOS 4.3.5-GA FastPath queued) with ESMTP id AVW32334; Thu, 26 Sep 2013 05:34:51 +0000 (GMT) Received: from LHREML404-HUB.china.huawei.com (10.201.5.218) by lhreml203-edg.huawei.com (172.18.7.221) with Microsoft SMTP Server (TLS) id 14.3.146.0; Thu, 26 Sep 2013 06:33:56 +0100 Received: from SZXEML448-HUB.china.huawei.com (10.82.67.191) by lhreml404-hub.china.huawei.com (10.201.5.218) with Microsoft SMTP Server (TLS) id 14.3.146.0; Thu, 26 Sep 2013 06:34:47 +0100 Received: from szxeml557-mbx.china.huawei.com ([169.254.5.217]) by szxeml448-hub.china.huawei.com ([10.82.67.191]) with mapi id 14.03.0146.000; Thu, 26 Sep 2013 13:34:40 +0800 From: h chan To: "Sri Gundavelli (sgundave)" , "dmm@ietf.org" Thread-Topic: [DMM] I-D Action: draft-ietf-dmm-requirements-07.txt Thread-Index: AQHOogjoa3/L9muNGkS5wozrKtEQEpnXRZkQgABo14A= Date: Thu, 26 Sep 2013 05:34:40 +0000 Message-ID: <6E31144C030982429702B11D6746B98C370CBC16@szxeml557-mbx.china.huawei.com> References: <1377486056.79906.YahooMailNeo@web163805.mail.gq1.yahoo.com> <24C0F3E22276D9438D6F366EB89FAEA8116A4D4F@xmb-aln-x03.cisco.com> Accept-Language: en-US, zh-CN Content-Language: en-US X-MS-Has-Attach: X-MS-TNEF-Correlator: x-originating-ip: [10.46.81.5] Content-Type: multipart/alternative; boundary="_000_6E31144C030982429702B11D6746B98C370CBC16szxeml557mbxchi_" MIME-Version: 1.0 X-CFilter-Loop: Reflected Subject: Re: [DMM] I-D Action: draft-ietf-dmm-requirements-07.txt X-BeenThere: dmm@ietf.org X-Mailman-Version: 2.1.12 Precedence: list List-Id: Distributed Mobility Management Working Group List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Thu, 26 Sep 2013 05:35:02 -0000 --_000_6E31144C030982429702B11D6746B98C370CBC16szxeml557mbxchi_ Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: quoted-printable Sri, Thanks for the comments. We are replying under different sections in separa= te emails. More emails will follow. ----------------------------------------------- We accept your comment on the abstract and have changed the abstract to the= following in version 08. Please check. Abstract This document defines the requirements for Distributed Mobility Management (DMM). The hierarchical structure in traditional wireless networks has led primarily to centralized deployment models. As some wireless networks are evolving away from the hierarchical structure, such as in moving the content delivery servers closer to the users, a distributed model for mobility management can be useful to them. The following have contributed to the revisions in draft 08: Pierrick, Dape= ng, and Jouni From: dmm-bounces@ietf.org [mailto:dmm-bounces= @ietf.org] On Behalf Of Sri Gundavelli (sgundave) Sent: Sunday, August 25, 2013 10:04 PM To: dmm@ietf.org Subject: Re: [DMM] I-D Action: draft-ietf-dmm-requirements-07.txt Please see inline for some comments. Regards Sri Abstract This document defines the requirements for Distributed Mobility Management (DMM) in IPv6 deployments. The hierarchical structure in traditional wireless networks has led to deployment models which are in practice centralized. Mobility management with logically centralized mobility anchoring in current mobile networks is prone to suboptimal routing and raises scalability issues. Such centralized functions can lead to single points of failure and inevitably introduce longer delays and higher signaling loads for network operations related to mobility management. The objective is to enhance mobility management in order to meet the primary goals in network evolution, i.e., improve scalability, avoid single points of failure, enable transparent mobility support to upper layers only when needed, and so on. Distributed mobility management must be secure and may co-exist with existing network deployments and end hosts. [Sri] We don't need to argue against centralized model to justify distribut= ed model. In the absence of proper data to compare on both the approaches, = we cannot have such text. The pros and cons extend to both the models. Ple= ase simply state distributed model can be useful in certain deployments and= hence there is interest for this work. 1. Introduction In the past decade a fair number of mobility protocols have been standardized [RFC6275] [RFC5944] [RFC5380] [RFC6301] [RFC5213]. Although the protocols differ in terms of functions and associated message formats, we can identify a few key common features: o a centralized mobility anchor providing global reachability and an always-on experience to the user; o extensions to the base protocols to optimize handover performance while users roam across wireless cells; and o extensions to enable the use of heterogeneous wireless interfaces for multi-mode terminals (e.g. smartphones). [Sri] Currently defined mobility protocols support handovers and multiple a= ccess technologies ? Not sure, how these two "common features" make the ca= se for DMM If the point is about centralized anchor's, you don't need the = other two points. The presence of the centralized mobility anchor allows a mobile node to remain reachable after it has moved to a different network. The anchor point, among other tasks, ensures connectivity by forwarding packets destined to, or sent from, the mobile node. In practice, most of the deployed architectures today have a small number of centralized anchors managing the traffic of millions of mobile nodes. Compared with a distributed approach, a centralized approach is likely to have several issues or limitations affecting performance and scalability, which require costly network engineering to resolve. [Sri] All 3G/4G systems are based on this model and they are running fine. = Again, we don't need to argue against centralized model to justify distribu= ted model. Please simply state distributed model can be useful in certain d= eployments and hence the motivation for this work. To optimize handovers from the perspective of mobile nodes, the base protocols have been extended to efficiently handle packet forwarding between the previous and new points of attachment. These extensions are necessary when applications have stringent requirements in terms of delay. Notions of localization and distribution of local agents have been introduced to reduce signaling overhead at the centralized routing anchor point [Paper-Distributed.Centralized.Mobility]. Unfortunately, today we witness difficulties in getting such protocols deployed, resulting in sub-optimal choices for the network Operators. [Sri] I assume this is about hierarchical models / Chaining ? What are thes= e "difficulties in getting such protocols deployed" ? Moreover, the availability of multiple-interface host and the possibility of using several network interfaces simultaneously have motivated the development of even more protocol extensions to add more capabilities to the mobility management protocol. In the end, deployment is further complicated with the multitude of extensions. [Sri] Not sure I follow. Mobile IP protocols are in general access agnostic= . Not sure, what are the extensions that we have added on access-basis. As an effective transport method for multimedia data delivery, IP multicast support, including optimizations, have been introduced but by "patching-up" procedure after completing the design of reference mobility protocol, leading to network inefficiency and non-optimal routing. [Sri] Multicast related extensions have nothing to do with multi-access sup= port/host's capability to support multiple interfaces. Can you clarify ? Th= is text is not clear. Chan (Ed.), et al. Expires February 3, 2014 [Page 4] Internet-Draft DMM-Reqs August 2013 Mobile users are, more than ever, consuming Internet content; such traffic imposes new requirements on mobile core networks for data traffic delivery. The presence of content providers closer to Internet Service Providers (ISP) network requires taking into account local Content Delivery Networks (CDNs) while providing mobility services. Moreover, when the traffic demand exceeds available capacity, service providers need to implement new strategies such as selective traffic offload (e.g. 3GPP work items LIPA/SIPTO [TS.23.401]) through alternative access networks (e.g. WLAN) [Paper- Mobile.Data.Offloading]. [Sri] Please add reference to IETF SIPTO doc, RFC6909, before 23.401 :) A gateway selection mechanism also takes the user proximity into account within EPC [TS.29303]. These mechanisms were not pursued in the past owing to charging and billing reasons. Assigning a gateway anchor node from a visited network in roaming scenario has until recently been done and are limited to voice services only. Charging and billing require solutions beyond the mobility protocol. Both traffic offloading and CDN mechanisms could benefit from the development of mobile architectures with fewer levels of routing hierarchy introduced into the data path by the mobility management system. This trend towards so-called "flat networks" works best for direct communications among peers in the same geographical area. Distributed mobility management in a truly flat mobile architecture would anchor the traffic closer to the point of attachment of the user. Today's mobile networks present service providers with new challenges. Mobility patterns indicate that mobile nodes often remain attached to the same point of attachment for considerable periods of time [Paper-Locating.User]. Specific IP mobility management support is not required for applications that launch and complete their sessions while the mobile node is connected to the same point of attachment. However, currently, IP mobility support is designed for always-on operation, maintaining all parameters of the context for each mobile subscriber for as long as they are connected to the network. This can result in a waste of resources and unnecessary= costs for the service provider. [Sri] Is the intent of this is text is about routing based approaches ? If a mobile is attached to the network, it does have some state at the anch= or. Also, if we consider the home link in mobility models, there is no stat= e for the mobile node when it is at home. In case of DMM, or with the curre= nt models, if the gateway selection is based on the MN's location and when = there is no node mobility, there should not be any state at the anchor. CDM= A Infrequent node mobility coupled with application intelligence suggest that mobility support could be provided selectively, thus reducing the amount of context maintained in the network. The distributed mobility management (DMM) charter addresses two complementary aspects of mobility management procedures: the distribution of mobility anchors towards a more flat network and the dynamic activation/deactivation of mobility protocol support as an enabler to distributed mobility management. [Sri] Not sure, I follow this point on Dynamic activation/de-activation. Ca= n you clarify. The former aims at positioning mobility anchors (e.g., HA, LMA) closer to the user; ideally, mobility agents could be collocated with the first-hop Chan (Ed.), et al. Expires February 3, 2014 [Page 5] Internet-Draft DMM-Reqs August 2013 router. The latter, facilitated by the distribution of mobility anchors, aims at identifying when mobility support must be activated and identifying sessions that do not require mobility management support -- thus reducing the amount of state information that must be maintained in various mobility agents of the mobile network. The key idea is that dynamic mobility management relaxes some of the constraints of previously-standardized mobility management solutions and, by doing so, it can avoid the unnecessary establishment of mechanisms to forward traffic from an old to a new mobility anchor. [Sri] The DMM model should not exclude the case of centralized anchor and d= istributed data plane. This is sub-case of DMM. This document compares distributed mobility management with centralized mobility management in Section 3. The problems that can be addressed with DMM are summarized in Section 4. The mandatory requirements as well as the optional requirements are given in Section 5. Finally, security considerations are discussed in Section 6. The problem statement and the use cases [I-D.yokota-dmm-scenario] can be found in [Paper-Distributed.Mobility.Review]. 2. Conventions used in this document 2.1. Terminology All the general mobility-related terms and their acronyms used in this document are to be interpreted as defined in the Mobile IPv6 base specification [RFC6275], in the Proxy mobile IPv6 specification [RFC5213], and in Mobility Related Terminology [RFC3753]. These terms include the following: mobile node (MN), correspondent node (CN), and home agent (HA) as per [RFC6275]; local mobility anchor (LMA) and mobile access gateway (MAG) as per [RFC5213], and context as per [RFC3753]. In addition, this draft introduces the following term. Mobility context is the collection of information required to provide mobility management support for a given mobile node. [Sri] There needs to be some definition of "centrally deployed mobility anc= hor". This term is used through-out the document. What is central, what is = local ? Even in the so called, "centralized anchor models", a gateway can b= e enabled locally. Ex: LMA/MAG, PGW/SGW functions can exist on the same nod= e. 3. Centralized versus distributed mobility management Mobility management functions may be implemented at different layers of the protocol stack. At the IP (network) layer, they may reside in the network or in the mobile node. In particular, a network-based solution resides in the network only. It therefore enables mobility Chan (Ed.), et al. Expires February 3, 2014 [Page 6] Internet-Draft DMM-Reqs August 2013 for existing hosts and network applications which are already in deployment but lack mobility support. [Sri] Above text is bit confusing to me, specially the last sentence. Mobil= ity management can be based on client-based, or network-based. At the IP layer, a mobility management protocol supporting session continuity is typically based on the principle of distinguishing between identifier and routing address and maintaining a mapping between the two. [Sri] Replace, "Session continuity" with "IP mobility" or "IP address conti= nuity". In Mobile IP, the home address serves as an identifier of the device whereas the care-of-address (CoA) takes the role of the routing address. The binding between these two is maintained at the home agent (mobility anchor). If packets can be continuously delivered to a mobile node at its home address, then all sessions using that home address are unaffected even though the routing address (CoA) changes. [Sri] We should leave it at, "IP address mobility". The next two subsections explain centralized and distributed mobility management functions in the network. 3.1. Centralized mobility management In centralized mobility management, the mapping information between the persistent node identifier and the locator IP address of a mobile node (MN) is kept at a single mobility anchor. At the same time, packets destined to the MN are routed via this anchor. [Sri] Can we use the MIP terminology, home address/Care-of address terminol= ogy, as supposed to LISP terminology ? In other words, such mobility management systems are centralized in both the control plane and the data plane (mobile node IP traffic). Many existing mobility management deployments make use of centralized mobility anchoring in a hierarchical network architecture, as shown in Figure 1. Examples of such centralized mobility anchors are the home agent (HA) and local mobility anchor (LMA) in Mobile IPv6 [RFC6275] and Proxy Mobile IPv6 [RFC5213], respectively. Current cellular networks such as the Third Generation Partnership Project (3GPP) GPRS networks, CDMA networks, and 3GPP Evolved Packet System (EPS) networks employ centralized mobility management too. In particular, the Gateway GPRS Support Node (GGSN), Serving GPRS Support Node (SGSN) and Radio Network Controller (RNC) in the 3GPP GPRS hierarchical network, and the Packet Data Network Gateway (P-GW) and Serving Gateway (S-GW) in the 3GPP EPS network all act as anchors in a hierarchy. Chan (Ed.), et al. Expires February 3, 2014 [Page 7] Internet-Draft DMM-Reqs August 2013 3G GPRS 3GPP EPS MIP/PMIP +------+ +------+ +------+ | GGSN | | P-GW | |HA/LMA| +------+ +------+ +------+ /\ /\ /\ / \ / \ / \ / \ / \ / \ / \ / \ / \ / \ / \ / \ / \ / \ / \ / \ / \ / \ +------+ +------+ +------+ +------+ +------+ +------+ | SGSN | | SGSN | | S-GW | | S-GW | |MN/MAG| |MN/MAG| +------+ +------+ +------+ +------+ +------+ +------+ /\ /\ / \ / \ / \ / \ +---+ +---+ +---+ +---+ |RNC| |RNC| |RNC| |RNC| +---+ +---+ +---+ +---+ Figure 1. Centralized mobility management. 3.2. Distributed mobility management Mobility management functions may also be distributed to multiple networks as shown in Figure 2, so that a mobile node in any of these networks may be served by a nearby mobility function (MF). +------+ +------+ +------+ +------+ | MF | | MF | | MF | | MF | +------+ +------+ +------+ +------+ | +----+ | MN | +----+ Figure 2. Distributed mobility management. Mobility management may be partially or fully distributed. In the former case only the data plane is distributed. Fully distributed mobility management implies that both the data plane and the control plane are distributed. Such concepts of data and control plane separation are not yet described in the IETF developed mobility protocols so far but are described in detail in [I-D.yokota-dmm- scenario]. While mobility management can be distributed, it is not necessary for other functions such as subscription management, Chan (Ed.), et al. Expires February 3, 2014 [Page 8] Internet-Draft DMM-Reqs August 2013 subscription database, and network access authentication to be similarly distributed. [Sri] The case of centralized CP and distributed DP is covered in IETF docs= , http://datatracker.ietf.org/doc/draft-wakikawa-netext-pmip-cp-up-separation= / This is a variant of the DMM models. A distributed mobility management scheme for flat IP-based mobile network architecture consisting of access nodes is proposed in [Paper-Distributed.Dynamic.Mobility]. Its benefits over centralized mobility management are shown through simulations in [Paper- Distributed.Centralized.Mobility]. Moreover, the (re)use and extension of existing protocols in the design of both fully distributed mobility management [Paper-Migrating.Home.Agents] [Paper- Distributed.Mobility.SAE] and partially distributed mobility management [Paper-Distributed.Mobility.PMIP] [Paper- Distributed.Mobility.MIP] have been reported in the literature. Therefore, before designing new mobility management protocols for a future flat IP architecture, it is recommended to first consider whether existing mobility management protocols can be extended to serve a flat IP architecture. [Sri] Lot of unnecessary text in this document. Not sure, we need all of th= is text. 4. Problem Statement The problems that can be addressed with DMM are summarized in the following: PS1: Non-optimal routes Routing via a centralized anchor often results in a longer route. [Sri] "longer route" ? I assume this is about routing/tx delay. Please re-= word. The problem is manifested, for example, when accessing a local server or servers of a Content Delivery Network (CDN), or when receiving locally available IP multicast or sending IP multicast packets. [Sri] Does RFC 6705, or RFC 6909 does not address this issue ? May be the C= DN example is incorrect. PS2: Divergence from other evolutionary trends in network architectures such as distribution of content delivery. Centralized mobility management can become non-optimal with a flat network architecture. [Sri] How is this making the case of DMM ? We want the MN to access content= locally in the access network and we want localized routing. We have that = in the form of 6705 and 6909. The other approach is give localized IP addre= sses and have the content locally accessed. There are simply two many consi= derations and points may be valid, but when we bring all those assumptions.= We are bringing these points in a less logical manner without stating the = assumptions. PS3: Low scalability of centralized tunnel management and mobility context maintenance Setting up tunnels through a central anchor and maintaining mobility context for each MN usually requires more concentrated resources in a centralized design, thus reducing scalability. Distributing the tunnel maintenance function and the mobility context maintenance function among different network entities with proper signaling protocol design can increase scalability. Chan (Ed.), et al. Expires February 3, 2014 [Page 9] Internet-Draft DMM-Reqs August 2013 PS4: Single point of failure and attack Centralized anchoring designs may be more vulnerable to single points of failures and attacks than a distributed system. The impact of a successful attack on a system with centralized mobility management can be far greater as well. PS5: Unnecessarily reserving resources to provide mobility support to nodes that do not need such support IP mobility support is not always required, and not every parameter of mobility context is always used. For example, some applications do not need a stable IP address during a handover to maintain session continuity. Sometimes, the entire application session runs while the terminal does not change the point of attachment. Besides, some sessions, e.g. SIP-based sessions, can handle mobility at the application layer and hence do not need IP mobility support; it is then more efficient to deactivate IP mobility support for such sessions. [Sri] Mobility systems today do support the aspect of service for the subs= criber, as "Simple IP", or "Mobile IP". A PDSN can assign a local IP addres= s and it does not have to be the home address. Network does have this intel= ligence, but what is missing is the client's ability to pick the correct ty= pe of IP address, among different IP addresses. The network is also the mis= sing the aspect of marking those addresses with proper properties. The curr= ently active drafts in IETF are to address this issue. We should talk about= these missing semantics. draft-bhandari-dhc-class-based-prefix-05 draft-korhonen-6man-prefix-properties-02 PS6: (Related problem) Mobility signaling overhead with peer-to-peer communication Wasting resources when mobility signaling (e.g., maintenance of the tunnel, keep alive signaling, etc.) is not turned off for peer-to-peer communication. Peer-to-peer communications have particular traffic patterns that often do not benefit from mobility support from the network. Thus, the associated mobility support signaling (e.g., maintenance of the tunnel, keep alive signaling, etc.) wastes network resources for no application gain. In such a case, it is better to enable mobility support selectively. [Sri] How is PS6 different from PS5 ? We talk about application's ability t= o pick the address with or without mobility properties in PS5. So, the traf= fic patterns can be localized based on the application requirements. But, e= ven if we take the argument that mobility is not required, operator needs v= isibility into these flows and so they can charge. Again, we have 6705 and= 6909 for adjusting to those traffic patterns. So, this P without those con= siderations is incomplete. PS7: (Related problem) Deployment with multiple mobility solutions There are already many variants and extensions of MIP. Deployment of new mobility management solutions can be challenging, and debugging difficult, when they must co-exist with solutions already in the field. PS8: Duplicate multicast traffic IP multicast distribution over architectures using IP mobility solutions (e.g. RFC6224) may lead to convergence of duplicated multicast subscriptions towards the downstream tunnel entity (e.g. MAG in PMIPv6). Concretely, when multicast subscription for individual mobile nodes is coupled with mobility tunnels (e.g. PMIPv6 tunnel), duplicate multicast subscription(s) is Chan (Ed.), et al. Expires February 3, 2014 [Page 10] Internet-Draft DMM-Reqs August 2013 prone to be received through different upstream paths. This problem may also exist or be more severe in a distributed mobility environment. [Sri] Is this for MN's from different LMA's attached to the same MAG ? 5. Requirements After comparing distributed mobility management against centralized deployment in Section 3, this section identifies the following requirements: 5.1. Distributed processing REQ1: Distributed processing IP mobility, network access and routing solutions provided by DMM MUST enable distributed processing for mobility management so that traffic does not need to traverse centrally deployed mobility anchors and thereby avoid non-optimal routes. Motivation: This requirement is motivated by current trends in network evolution: (a) it is cost- and resource-effective to cache and distribute content by combining distributed mobility anchors with caching systems (e.g., CDN); (b) the significantly larger number of mobile nodes and flows call for improved scalability; (c) single points of failure are avoided in a distributed system; (d) threats against centrally deployed anchors, e.g., home agent and local mobility anchor, are mitigated in a distributed system. This requirement addresses the problems PS1, PS2, PS3, and PS4 described in Section 4. (Existing route optimization is only a host- based solution. On the other hand, localized routing with PMIPv6 addresses only a part of the problem where both the MN and the CN are located in the PMIP domain and attached to a MAG, and is not applicable when the CN is outside the PMIP domain.) [Sri] I'm still stuck on the CDN example driving this requirement. 5.2. Transparency to Upper Layers when needed REQ2: Transparency to Upper Layers when needed DMM solutions MUST provide transparent mobility support above the IP layer when needed. Such transparency is needed, for example, when, upon change of point of attachment to the network, an application flow cannot cope with a change in the IP address. However, it is not always necessary to maintain a stable home IP address or prefix for every application or at all times for a mobile node. [Sri] Please reflect the two key aspects of this requirement: * Network can assign IP addresses with different properties; It carr= ies those properties * Applications have different requirements and will pick the address= with the correct property. Chan (Ed.), et al. Expires February 3, 2014 [Page 11] Internet-Draft DMM-Reqs August 2013 Motivation: The motivation of this requirement is to enable more efficient use of network resources and more efficient routing by not maintaining context at the mobility anchor when there is no such need. This requirement addresses the problem PS5 as well as the related problem PS6 stated in Section 4. 5.3. IPv6 deployment REQ3: IPv6 deployment DMM solutions SHOULD target IPv6 as the primary deployment environment and SHOULD NOT be tailored specifically to support IPv4, in particular in situations where private IPv4 addresses and/or NATs are used. Motivation: This requirement conforms to the general orientation of IETF work. DMM deployment is foreseen in mid- to long-term horizon, when IPv6 is expected to be far more common than today. This requirement avoids the unnecessarily complexity in solving the problems in Section 4 for IPv4, which will not be able to use some of the IPv6-specific features. 5.4. Existing mobility protocols REQ4: Existing mobility protocols A DMM solution SHOULD first consider reusing and extending IETF-standardized protocols before specifying new protocols. Motivation: Reuse of existing IETF work is more efficient and less error-prone. This requirement attempts to avoid the need of new protocols development and therefore their potential problems of being time- consuming and error-prone. 5.5. Co-existence REQ5: Co-existence with deployed networks and hosts The DMM solution MUST be able to co-exist with existing network deployments and end hosts. For example, depending on the environment in which DMM is deployed, DMM solutions may need to be compatible with other deployed mobility protocols Chan (Ed.), et al. Expires February 3, 2014 [Page 12] Internet-Draft DMM-Reqs August 2013 or may need to co-exist with a network or mobile hosts/routers that do not support DMM protocols. The mobile node may also move between different access networks, where some of them may support neither DMM nor another mobility protocol. Furthermore, a DMM solution SHOULD work across different networks, possibly operated as separate administrative domains, when allowed by the trust relationship between them. Motivation: (a) to preserve backwards compatibility so that existing networks and hosts are not affected and continue to function as usual, and (b) enable inter-domain operation if desired. This requirement addresses the related problem PS7 described in Section 4. 5.6. Security considerations REQ6: Security considerations A DMM solution MUST not introduce new security risks or amplify existing security risks against which the existing security mechanisms/protocols cannot offer sufficient protection. Motivation: Various attacks such as impersonation, denial of service, man-in-the-middle attacks, and so on, may be launched in a DMM deployment. For instance, an illegitimate node may attempt to access a network providing DMM. Another example is that a malicious node can forge a number of signaling messages thus redirecting traffic from its legitimate path. Consequently, the specific node is under a denial of service attack, whereas other nodes do not receive their traffic. Accordingly, security mechanisms/protocols providing access control, integrity, authentication, authorization, confidentiality, etc. can be used to protect the DMM entities as they are already used to protect against existing networks and existing mobility protocols defined in IETF. In addition, end-to-end security measures between communicating nodes may already be used when deploying existing mobility protocols where the signaling messages travel over the Internet. For instance, EAP-based authentication can be used for network access security, while IPsec can be used for end-to-end security. When the existing security mechanisms/protocols are applied to protect the DMM entities, the security risks that may be introduced by DMM MUST be considered to be eliminated. Else the security protection would be degraded in the DMM solution versus in existing mobility protocols. Chan (Ed.), et al. Expires February 3, 2014 [Page 13] Internet-Draft DMM-Reqs August 2013 This requirement prevents a DMM solution from introducing uncontrollable problems of potentially insecure mobility management protocols which make deployment infeasible because platforms conforming to the protocols are at risk for data loss and numerous other dangers, including financial harm to the users. 5.7. Multicast REQ7: Multicast considerations DMM SHOULD consider multicast early so that solutions can be developed not only to provide IP mobility support when it is needed, but also to avoid network inefficiency issues in multicast traffic delivery (such as duplicate multicast subscriptions towards the downstream tunnel entities). The multicast solutions should therefore avoid restricting the management of all IP multicast traffic to a single host through a dedicated (tunnel) interface on multicast-capable access routers. Motivation: Existing multicast deployment have been introduced after completing the design of the reference mobility protocol, then optimization and extensions have been followed by "patching-up" procedure, thus leading to network inefficiency and non-optimal routing. The multicast solutions should therefore be required to consider efficiency nature in multicast traffic delivery. This requirement addresses the problems PS1 and PS8 described in Section 4. 6. Security Considerations Please refer to the discussion under Security requirement in Section 5.6. 7. IANA Considerations None 8. Co-authors and Contributors This problem statement document is a joint effort among the numerous participants. Each individual has made significant contributions to this work and have been listed as co-authors. Chan (Ed.), et al. Expires February 3, 2014 [Page 14] Internet-Draft DMM-Reqs August 2013 9. References 9.1. Normative References [RFC2119] Bradner, S., "Key words for use in RFCs to Indicate Requirement Levels", BCP 14, RFC 2119, March 1997. 9.2. Informative References [I-D.yokota-dmm-scenario] Yokota, H., Seite, P., Demaria, E., and Z. Cao, "Use case scenarios for Distributed Mobility Management", draft-yokota-dmm-scenario-00 (work in progress), October 2010. [Paper-Distributed.Centralized.Mobility] Bertin, P., Bonjour, S., and J-M. Bonnin, "A Distributed or Centralized Mobility", Proceedings of Global Communications Conference (GlobeCom), December 2009. [Paper-Distributed.Dynamic.Mobility] Bertin, P., Bonjour, S., and J-M. Bonnin, "A Distributed Dynamic Mobility Management Scheme Designed for Flat IP Architectures", Proceedings of 3rd International Conference on New Technologies, Mobility and Security (NTMS), 2008. [Paper-Distributed.Mobility.MIP] Chan, H., "Distributed Mobility Management with Mobile IP", Proceedings of IEEE International Communication Conference (ICC) Workshop on Telecommunications: from Research to Standards, June 2012. [Paper-Distributed.Mobility.PMIP] Chan, H., "Proxy Mobile IP with Distributed Mobility Anchors", Proceedings of GlobeCom Workshop on Seamless Wireless Mobility, December 2010. [Paper-Distributed.Mobility.Review] Chan, H., Yokota, H., Xie, J., Seite, P., and D. Liu, "Distributed and Dynamic Mobility Management in Mobile Internet: Current Approaches and Issues, Journal of Communications, vol. 6, no. 1, pp. 4-15, Feb 2011.", Proceedings of GlobeCom Workshop on Seamless Wireless Mobility, February 2011. [Paper-Distributed.Mobility.SAE] Fisher, M., Anderson, F., Kopsel, A., Schafer, G., and M. Chan (Ed.), et al. Expires February 3, 2014 [Page 15] Internet-Draft DMM-Reqs August 2013 Schlager, "A Distributed IP Mobility Approach for 3G SAE", Proceedings of the 19th International Symposium on Personal, Indoor and Mobile Radio Communications (PIMRC), 2008. [Paper-Locating.User] Kirby, G., "Locating the User", Communication International, 1995. [Paper-Migrating.Home.Agents] Wakikawa, R., Valadon, G., and J. Murai, "Migrating Home Agents Towards Internet-scale Mobility Deployments", Proceedings of the ACM 2nd CoNEXT Conference on Future Networking Technologies, December 2006. [Paper-Mobile.Data.Offloading] Lee, K., Lee, J., Yi, Y., Rhee, I., and S. Chong, "Mobile Data Offloading: How Much Can WiFi Deliver?", SIGCOMM 2010, 2010. [RFC3753] Manner, J. and M. Kojo, "Mobility Related Terminology", RFC 3753, June 2004. [RFC5213] Gundavelli, S., Leung, K., Devarapalli, V., Chowdhury, K., and B. Patil, "Proxy Mobile IPv6", RFC 5213, August 2008. [RFC5380] Soliman, H., Castelluccia, C., ElMalki, K., and L. Bellier, "Hierarchical Mobile IPv6 (HMIPv6) Mobility Management", RFC 5380, October 2008. [RFC5944] Perkins, C., "IP Mobility Support for IPv4, Revised", RFC 5944, November 2010. [RFC6275] Perkins, C., Johnson, D., and J. Arkko, "Mobility Support in IPv6", RFC 6275, July 2011. [RFC6301] Zhu, Z., Wakikawa, R., and L. Zhang, "A Survey of Mobility Support in the Internet", RFC 6301, July 2011. [TS.23.401] 3GPP, "General Packet Radio Service (GPRS) enhancements for Evolved Universal Terrestrial Radio Access Network (E-UTRAN) access", 3GPP TR 23.401 10.10.0, March 2013. [TS.29303] 3GPP, "Domain Name System Procedures; Stage 3", 3GPP TR 23.303 11.2.0, September 2012. Chan (Ed.), et al. Expires February 3, 2014 [Page 16] Internet-Draft DMM-Reqs August 2013 Authors' Addresses H Anthony Chan (editor) Huawei Technologies (more co-authors on P. 17) 5340 Legacy Dr. Building 3, Plano, TX 75024, USA Email: h.a.chan@ieee.org Dapeng Liu China Mobile Unit2, 28 Xuanwumenxi Ave, Xuanwu District, Beijing 100053, China Email: liudapeng@chinamobile.com Pierrick Seite Orange 4, rue du Clos Courtel, BP 91226, Cesson-Sevigne 35512, France Email: pierrick.seite@orange.com Hidetoshi Yokota KDDI Lab 2-1-15 Ohara, Fujimino, Saitama, 356-8502 Japan Email: yokota@kddilabs.jp Jouni Korhonen Nokia Siemens Networks Email: jouni.korhonen@nsn.com - Charles E. Perkins Huawei Technologies Email: charliep@computer.org - Melia Telemaco Alcatel-Lucent Bell Labs Email: telemaco.melia@alcatel-lucent.com - Elena Demaria Telecom Italia via G. Reiss Romoli, 274, TORINO, 10148, Italy Email: elena.demaria@telecomitalia.it - Jong-Hyouk Lee RSM Department, Telecom Bretagne Cesson-Sevigne, 35512, France Email: jh.lee@telecom-bretagne.eu - Chan (Ed.), et al. Expires February 3, 2014 [Page 17] Internet-Draft DMM-Reqs August 2013 Kostas Pentikousis Huawei Technologies Carnotstr. 4 10587 Berlin, Germany Email: k.pentikousis@huawei.com - Tricci So ZTE Email: tso@zteusa.com - Carlos J. Bernardos Universidad Carlos III de Madrid Av. Universidad, 30, Leganes, Madrid 28911, Spain Email: cjbc@it.uc3m.es - Peter McCann Huawei Technologies Email: PeterMcCann@huawei.com - Seok Joo Koh Kyungpook National University, Korea Email: sjkoh@knu.ac.kr - Wen Luo ZTE No.68, Zijinhua RD,Yuhuatai District, Nanjing, Jiangsu 210012, China Email: luo.wen@zte.com.cn - Sri Gundavelli sgundave@cisco.com - Marco Liebsch NEC Laboratories Europe Email: liebsch@neclab.eu - Carl Williams MCSR Labs Email: carlw@mcsr-labs.org - Seil Jeon Instituto de Telecomunicacoes, Aveiro Email: seiljeon@av.it.pt - Sergio Figueiredo Universidade de Aveiro Email: sfigueiredo@av.it.pt - Stig Venaas Email: stig@venaas.com Chan (Ed.), et al. Expires February 3, 2014 [Page 18] Internet-Draft DMM-Reqs August 2013 - Luis Miguel Contreras Murillo Email: lmcm@tid.es - Juan Carlos Zuniga Email: JuanCarlos.Zuniga@InterDigital.com - Alexandru Petrescu Email: alexandru.petrescu@gmail.com - Georgios Karagiannis Email: g.karagiannis@utwente.nl - Julien Laganier jlaganier@juniper.net - Wassim Michel Haddad Wassam.Haddad@ericsson.com - Dirk von Hugo Dirk.von-Hugo@telekom.de - Ahmad Muhanna amuhanna@awardsolutions.com - Byoung-Jo Kim ATT Labs macsbug@research.att.com - Hassan Aliahmad Orange hassan.aliahmad@orange.com - Chan (Ed.), et al. Expires February 3, 2014 [Page 19] --_000_6E31144C030982429702B11D6746B98C370CBC16szxeml557mbxchi_ Content-Type: text/html; charset="us-ascii" Content-Transfer-Encoding: quoted-printable

Sri,

 <= /p>

Thanks for the comments. = We are replying under d= ifferent sections in separate emails. More emails will follow.

-------------------------= ----------------------

We accept your comment on the abstract and have changed in version= 08. Please check.

 <= /p>

Abstract

 

   This document defines the requirements for Di= stributed Mobility

   Management (DMM).  The hierarchical stru= cture in traditional wireless

   networks has led primarily to centralized dep= loyment models.  As some

   wireless networks are evolving away from the = hierarchical structure,

   such as in moving the content delivery server= s closer to the users, a

   distributed model for mobility management can= be useful to them.

 <= /p>

The following have contri= buted to the revisions in draft 08: Pierrick, Dapeng, and Jouni

 <= /p>

 <= /p>

From: dmm-bounces@ietf.org [mailto:dmm-bounces@ietf.org] On Behalf Of Sri Gundavelli (sgundave)
Sent: Sunday, August 25, 2013 10:04 PM
To: dmm@ietf.org
Subject: Re: [DMM] I-D Action: draft-ietf-dmm-requirements-07.txt

 

Please see inline for some comments.

 

Regards

Sri

 

 
 
Abstract=

 
   This d=
ocument defines the requirements for Distributed Mobility=

   Manage=
ment (DMM) in IPv6 deployments.  The hierarchical structure in
   tradit=
ional wireless networks has led to deployment models which are
   in pra=
ctice centralized.  Mobility management with logically
   centra=
lized mobility anchoring in current mobile networks is prone to<=
/span>
   subopt=
imal routing and raises scalability issues.  Such centralized
   functi=
ons can lead to single points of failure and inevitably

   introd=
uce longer delays and higher signaling loads for network<=
/pre>

   operat=
ions related to mobility management.  The objective is to


   enhanc=
e mobility management in order to meet the primary goals in


   networ=
k evolution, i.e., improve scalability, avoid single points of


   failur=
e, enable transparent mobility support to upper layers only


   when n=
eeded, and so on.  Distributed mobility management must be<=
/span>


   secure=
 and may co-exist with existing network deployments and end


   hosts.=



[Sri] We don̵=
7;t need to argue against centralized model to justify distributed model. I=
n the absence of proper data to compare on both the approaches, we cannot h=
ave such text.  The pros and cons extend to both the models. Please si=
mply state distributed model can be useful in certain deployments and hence=
 there is interest for this work.
 
 
 
1.  Introducti=
on
 
   In the=
 past decade a fair number of mobility protocols have been
   standa=
rdized [RFC6275] [RFC5944] [RFC5380] [RFC6301] [RFC5213].=

   Althou=
gh the protocols differ in terms of functions and associated
   messag=
e formats, we can identify a few key common features:

 


   o =
; a centralized mobility anchor providing global reachability and an


   &=
nbsp;  always-on experience to the user;


 


   o =
; extensions to the base protocols to optimize handover performance


   &=
nbsp;  while users roam across wireless cells; and

 


   o =
; extensions to enable the use of heterogeneous wireless interfaces


   &=
nbsp;  for multi-mode terminals (e.g. smartphones).<=
/pre>

 


[Sri] Currently defined mobility =
protocols support handovers and multiple access technologies  ? Not su=
re, how these two "common features" make the case for DMM  <=
/span>If the point is about centralized ancho=
r's, you don't need the other two points. 


 


   The pr=
esence of the centralized mobility anchor allows a mobile node


   to rem=
ain reachable after it has moved to a different network.  The


   anchor=
 point, among other tasks, ensures connectivity by forwarding


   packet=
s destined to, or sent from, the mobile node.  In practice,=



   most o=
f the deployed architectures today have a small number of=



   centra=
lized anchors managing the traffic of millions of mobile nodes.<=
/span>


   Compar=
ed with a distributed approach, a centralized approach is=



   likely=
 to have several issues or limitations affecting performance


   and sc=
alability, which require costly network engineering to resolve.<=
/span>


 


[Sri] All 3G/4G s=
ystems are based on this model and they are running fine. Again, we donR=
17;t need to argue against centralized model to justify distributed model. =
Please simply state distributed model can be useful in certain deployments =
and hence the motivation for this work.




<=
o:p> 




   To opt=
imize handovers from the perspective of mobile nodes, the base


   protoc=
ols have been extended to efficiently handle packet forwarding


   betwee=
n the previous and new points of attachment.  These extensions


   are ne=
cessary when applications have stringent requirements in terms


   of del=
ay.  Notions of localization and distribution of local agents


   have b=
een introduced to reduce signaling overhead at the centralized


   routin=
g anchor point [Paper-Distributed.Centralized.Mobility].<=
/pre>

   Unfort=
unately, today we witness difficulties in getting such

   protoc=
ols deployed, resulting in sub-optimal choices for the network


   Operat=
ors.


[Sri] I assume th=
is is about hierarchical models / Chaining ? What are these "difficult=
ies in getting such protocols deployed" ?   


   M=
oreover, the availability of multiple-interface host and the


   possib=
ility of using several network interfaces simultaneously have


   motiva=
ted the development of even more protocol extensions to add


   more c=
apabilities to the mobility management protocol.  In the end,


   deploy=
ment is further complicated with the multitude of extensions.


[Sri] Not sure I follow. Mobile I=
P protocols are in general access agnostic. Not sure, what are the extensio=
ns that we have added on access-basis.=



 


 


   As an =
effective transport method for multimedia data delivery, IP


   multic=
ast support, including optimizations, have been introduced but


   by &qu=
ot;patching-up" procedure after completing the design of reference


   mobili=
ty protocol, leading to network inefficiency and non-optimal


   routin=
g.


 


[Sri] Multicast r=
elated extensions have nothing to do with multi-access support/host's capab=
ility to support multiple interfaces. Can you clarify ? This text is not cl=
ear.


Chan (Ed.), et al.&=
nbsp;     Expires February 3, 2014   &nb=
sp;            [Page=
 4]


 


Internet-Draft =
;            &n=
bsp;    DMM-Reqs        &=
nbsp;           &nbs=
p;August 2013


 


 


   Mobile=
 users are, more than ever, consuming Internet content; such


   traffi=
c imposes new requirements on mobile core networks for data


   traffi=
c delivery.  The presence of content providers closer to


   Intern=
et Service Providers (ISP) network requires taking into account<=
/span>


   local =
Content Delivery Networks (CDNs) while providing mobility=



   servic=
es.  Moreover, when the traffic demand exceeds available


   capaci=
ty, service providers need to implement new strategies such as


   select=
ive traffic offload (e.g. 3GPP work items LIPA/SIPTO

   [TS.23=
.401]) through alternative access networks (e.g.  WLAN) [Paper-


   Mobile=
.Data.Offloading].  


[Sri] Please add =
reference to IETF SIPTO doc, RFC6909, before 23.401 :)


 


A gateway selection=
 mechanism also takes


   the us=
er proximity into account within EPC [TS.29303].  These


   mechan=
isms were not pursued in the past owing to charging and billing<=
/span>


   reason=
s.   Assigning a gateway anchor node from a visited network in


   roamin=
g scenario has until recently been done and are limited to


   voice =
services only.  Charging and billing require solutions beyond


   the mo=
bility protocol.


 


   Both t=
raffic offloading and CDN mechanisms could benefit from the


   develo=
pment of mobile architectures with fewer levels of routing


   hierar=
chy introduced into the data path by the mobility management


   system=
.  This trend towards so-called "flat networks" works best f=
or


   direct=
 communications among peers in the same geographical area.


   Distri=
buted mobility management in a truly flat mobile architecture


   would =
anchor the traffic closer to the point of attachment of the


   user.<=
o:p>


 


   Today'=
s mobile networks present service providers with new

   challe=
nges.  Mobility patterns indicate that mobile nodes often


   remain=
 attached to the same point of attachment for considerable


   period=
s of time [Paper-Locating.User].  Specific IP mobility


   manage=
ment support is not required for applications that launch and


   comple=
te their sessions while the mobile node is connected to the


   same p=
oint of attachment. 


 However, currently=
, IP mobility support is


   design=
ed for always-on operation, maintaining all parameters of the


   contex=
t for each mobile subscriber for as long as they are connected


   to the=
 network.  This can result in a waste of resources and unnecessar=
y costs for the service provider. 


[Sri]  Is th=
e intent of this is text is about routing based approaches ?


If a mobile is at=
tached to the network, it does have some state at the anchor. Also, if we c=
onsider the home link in mobility models, there is no state for the mobile =
node when it is at home. In case of DMM, or with the current models, if the=
 gateway selection is based on the MN's location and when there is no node =
mobility, there should not be any state at the anchor. CDMA


 


 


 Infrequent node mo=
bility


   couple=
d with application intelligence suggest that mobility support


   could =
be provided selectively, thus reducing the amount of context


   mainta=
ined in the network.


 


   The di=
stributed mobility management (DMM) charter addresses two=



   comple=
mentary aspects of mobility management procedures: the

   distri=
bution of mobility anchors towards a more flat network and the


   dynami=
c activation/deactivation of mobility protocol support as an


   enable=
r to distributed mobility management.  


[Sri] Not sure, I=
 follow this point on Dynamic activation/de-activation. Can you clarify.


 


The former aims at<=
o:p>


   positi=
oning mobility anchors (e.g., HA, LMA) closer to the user;


   ideall=
y, mobility agents could be collocated with the first-hop=



 


 


 


Chan (Ed.), et al.&=
nbsp;     Expires February 3, 2014   &nb=
sp;            [Page=
 5]


 


Internet-Draft =
;            &n=
bsp;    DMM-Reqs       &n=
bsp;            =
; August 2013


 


 


   router=
.  The latter, facilitated by the distribution of mobility<=
/span>


   anchor=
s, aims at identifying when mobility support must be activated


   and id=
entifying sessions that do not require mobility management


   suppor=
t -- thus reducing the amount of state information that must be<=
/span>


   mainta=
ined in various mobility agents of the mobile network.  The key


   idea i=
s that dynamic mobility management relaxes some of the

   constr=
aints of previously-standardized mobility management solutions


   and, b=
y doing so, it can avoid the unnecessary establishment of=



   mechan=
isms to forward traffic from an old to a new mobility anchor.


 


[Sri] The DMM mod=
el should not exclude the case of centralized anchor and distributed data p=
lane. This is sub-case of DMM. 


 


   This d=
ocument compares distributed mobility management with

   centra=
lized mobility management in Section 3.  The problems that can


   be add=
ressed with DMM are summarized in Section 4.  The mandatory=



   requir=
ements as well as the optional requirements are given in<=
/pre>

   Sectio=
n 5.  Finally, security considerations are discussed in Section


   6.


 


   The pr=
oblem statement and the use cases [I-D.yokota-dmm-scenario] can<=
/span>


   be fou=
nd in [Paper-Distributed.Mobility.Review].


 


 


2.  Convention=
s used in this document


 


2.1.  Terminol=
ogy


 


   All th=
e general mobility-related terms and their acronyms used in


   this d=
ocument are to be interpreted as defined in the Mobile IPv6


   base s=
pecification [RFC6275], in the Proxy mobile IPv6 specification


   [RFC52=
13], and in Mobility Related Terminology [RFC3753].  These<=
/span>


   terms =
include the following: mobile node (MN), correspondent node


   (CN), =
and home agent (HA) as per [RFC6275]; local mobility anchor


   (LMA) =
and mobile access gateway (MAG) as per [RFC5213], and context


   as per=
 [RFC3753].


 


   In add=
ition, this draft introduces the following term.


 


   Mobili=
ty context


 


   &=
nbsp;  is the collection of information required to provide mobility


   &=
nbsp;  management support for a given mobile node.

 


 


[Sri] There needs=
 to be some definition of "centrally deployed mobility anchor". T=
his term is used through-out the document. What is central, what is local ?=
 Even in the so called, "centralized anchor models", a gateway ca=
n be enabled locally. Ex: LMA/MAG, PGW/SGW functions can exist on the same =
node. 


 


3.  Centralize=
d versus distributed mobility management


 


   Mobili=
ty management functions may be implemented at different layers


   of the=
 protocol stack.  At the IP (network) layer, they may reside in


   the ne=
twork or in the mobile node.  In particular, a network-based


   soluti=
on resides in the network only.  It therefore enables mobility


 


 


 


Chan (Ed.), et al.&=
nbsp;     Expires February 3, 2014   &nb=
sp;            [Page=
 6]


 


Internet-Draft =
;            &n=
bsp;    DMM-Reqs        &=
nbsp;           &nbs=
p;August 2013


 


 


   for ex=
isting hosts and network applications which are already in


   deploy=
ment but lack mobility support.


 


[Sri] Above text =
is bit confusing to me, specially the last sentence. Mobility management ca=
n be based on client-based, or network-based. 


 


   At the=
 IP layer, a mobility management protocol supporting session


   cont=
inuity is typically based on the principle o=
f distinguishing


   betwee=
n identifier and routing address and maintaining a mapping


   betwee=
n the two.  


[Sri] Replace, &q=
uot;Session continuity" with "IP mobility" or "IP addre=
ss continuity". 


In Mobile IP, the h=
ome address serves as an


   identi=
fier of the device whereas the care-of-address (CoA) takes the


   role o=
f the routing address.  The binding between these two is


   mainta=
ined at the home agent (mobility anchor).  If packets can be


   contin=
uously delivered to a mobile node at its home address, then all<=
/span>


   sessio=
ns using that home address are unaffected even though the=



   routin=
g address (CoA) changes.


 


[Sri] We should l=
eave it at, "IP address mobility".


   The ne=
xt two subsections explain centralized and distributed mobility<=
/span>


   manage=
ment functions in the network.


 


3.1.  Centrali=
zed mobility management


 


   In cen=
tralized mobility management, the mapping information between


   the pe=
rsistent node identifier and the locator IP address of a mobile<=
/span>


   node (=
MN) is kept at a single mobility anchor.  At the same time,=



   packet=
s destined to the MN are routed via this anchor.


[Sri] Can we use =
the MIP terminology, home address/Care-of address terminology, as supposed =
to LISP terminology ? 


 


  In other


   words,=
 such mobility management systems are centralized in both the


   contro=
l plane and the data plane (mobile node IP traffic).

 


   Many e=
xisting mobility management deployments make use of centralized<=
/span>


   mobili=
ty anchoring in a hierarchical network architecture, as shown


   in Fig=
ure 1.  Examples of such centralized mobility anchors are the


   home a=
gent (HA) and local mobility anchor (LMA) in Mobile IPv6<=
/pre>

   [RFC62=
75] and Proxy Mobile IPv6 [RFC5213], respectively.  Current=



   cellul=
ar networks such as the Third Generation Partnership Project


   (3GPP)=
 GPRS networks, CDMA networks, and 3GPP Evolved Packet System


   (EPS) =
networks employ centralized mobility management too.  In


   partic=
ular, the Gateway GPRS Support Node (GGSN), Serving GPRS<=
/pre>

   Suppor=
t Node (SGSN) and Radio Network Controller (RNC) in the 3GPP


   GPRS h=
ierarchical network, and the Packet Data Network Gateway (P-GW)<=
/span>


   and Se=
rving Gateway (S-GW) in the 3GPP EPS network all act as anchors<=
/span>


   in a h=
ierarchy.


 


 


 


 


 


 


 


 


 


 


 


 


Chan (Ed.), et al.&=
nbsp;     Expires February 3, 2014   &nb=
sp;            [Page=
 7]


 


Internet-Draft =
;            &n=
bsp;    DMM-Reqs       &n=
bsp;            =
; August 2013


 


 


   &=
nbsp;     3G GPRS      &n=
bsp;          3GPP EPS &n=
bsp;            =
;  MIP/PMIP


   &=
nbsp;     +------+     =
;           +------&#=
43;            =
    +------+


   &=
nbsp;     | GGSN |      &=
nbsp;         | P-GW |  &=
nbsp;           &nbs=
p; |HA/LMA|


   &=
nbsp;     +------+     =
;           +------&#=
43;            =
    +------+


   &=
nbsp;        /\    &=
nbsp;            &nb=
sp;    /\       &nbs=
p;            &=
nbsp; /\


   &=
nbsp;       /  \    =
            &nb=
sp;   /  \        &n=
bsp;           /  \<=
o:p>


   &=
nbsp;      /    \   =
            &nb=
sp;  /    \       &n=
bsp;          /  &nb=
sp; \


   &=
nbsp;     /      \  =
            &nb=
sp; /      \      &n=
bsp;         /   &nb=
sp;  \


   &=
nbsp;    /        \ =
             /&=
nbsp;       \     &n=
bsp;        /    &nb=
sp;   \


   &=
nbsp;   /          \=
            / &=
nbsp;        \    &n=
bsp;       /     &nb=
sp;    \


   &=
nbsp;  /          &n=
bsp; \          /  &=
nbsp;         \   &n=
bsp;      /      &nb=
sp;     \


  +------&=
#43;      +------+  +------+&=
nbsp;     +------+  +------+ =
     +------+


  | SGSN |&nbs=
p;     | SGSN |  | S-GW |    &=
nbsp; | S-GW |  |MN/MAG|      |MN/MAG|


  +------&=
#43;      +------+  +------+&=
nbsp;     +------+  +------+ =
     +------+


   &=
nbsp; /\            =
/\


    =
/  \          /  \


   / =
;   \        /  &nbs=
p; \


+---+ =
 +---+  +---+  +---+

|RNC|  |RNC|&n=
bsp; |RNC|  |RNC|


+---+ =
 +---+  +---+  +---+

 


   Figure=
 1.  Centralized mobility management.


 


3.2.  Distribu=
ted mobility management


 


   Mobili=
ty management functions may also be distributed to multiple


   networ=
ks as shown in Figure 2, so that a mobile node in any of these


   networ=
ks may be served by a nearby mobility function (MF).

 


 


   &=
nbsp;           &nbs=
p;    +------+  +------+  +---=
---+  +------+


   &=
nbsp;           &nbs=
p;    |  MF  |  |  MF  |  |&nb=
sp; MF  |  |  MF  |


   &=
nbsp;           &nbs=
p;    +------+  +------+  +---=
---+  +------+


    =
            &nb=
sp;            =
            &nb=
sp; |


   &=
nbsp;           &nbs=
p;            &=
nbsp;            =
3;----+


   &=
nbsp;           &nbs=
p;            &=
nbsp;            | M=
N |


   &=
nbsp;           &nbs=
p;            &=
nbsp;            =
3;----+


 


   Figure=
 2.  Distributed mobility management.


 


 


   Mobili=
ty management may be partially or fully distributed.  In the


   former=
 case only the data plane is distributed.  Fully distributed


   mobili=
ty management implies that both the data plane and the control


   plane =
are distributed.  Such concepts of data and control plane


   separa=
tion are not yet described in the IETF developed mobility=



   protoc=
ols so far but are described in detail in [I-D.yokota-dmm-


   scenar=
io].  While mobility management can be distributed, it is not


   necess=
ary for other functions such as subscription management,<=
/pre>

 


 


 


Chan (Ed.), et al.&=
nbsp;     Expires February 3, 2014   &nb=
sp;            [Page=
 8]


 


Internet-Draft =
;            &n=
bsp;    DMM-Reqs       &n=
bsp;            =
; August 2013


 


 


   subscr=
iption database, and network access authentication to be<=
/pre>

   simila=
rly distributed.


 


[Sri] The case of centralized CP =
and distributed DP is covered in IETF docs, 


ht=
tp://datatracker.ietf.org/doc/draft-wakikawa-netext-pmip-cp-up-separation/<=
/a>


This is a varian=
t of the DMM models. <=
/pre>

 


   A dist=
ributed mobility management scheme for flat IP-based mobile


   networ=
k architecture consisting of access nodes is proposed in<=
/pre>

   [Paper=
-Distributed.Dynamic.Mobility].  Its benefits over centralized


   mobili=
ty management are shown through simulations in [Paper-

   Distri=
buted.Centralized.Mobility].  Moreover, the (re)use and


   extens=
ion of existing protocols in the design of both fully

   distri=
buted mobility management [Paper-Migrating.Home.Agents] [Paper-<=
/span>


   Distri=
buted.Mobility.SAE] and partially distributed mobility

   manage=
ment [Paper-Distributed.Mobility.PMIP] [Paper-


   Distri=
buted.Mobility.MIP] have been reported in the literature.=



   Theref=
ore, before designing new mobility management protocols for a


   future=
 flat IP architecture, it is recommended to first consider


   whethe=
r existing mobility management protocols can be extended to


   serve =
a flat IP architecture.


 


[Sri] =
Lot of unnecessary text in this document. Not sure, we need all o=
f this text.


 


 


4.  Problem St=
atement


 


   The pr=
oblems that can be addressed with DMM are summarized in the


   follow=
ing:


 


   PS1:&n=
bsp; Non-optimal routes


 


   &=
nbsp;     Routing via a centralized anchor often result=
s in a longer


   &=
nbsp;     route.  


[Sri] "longer route" ?&=
nbsp; I assume this is about routing/tx delay. Please re-word.


 


The problem is mani=
fested, for example, when accessing


   &=
nbsp;     a local server or servers of a Content Delive=
ry Network (CDN),


   &=
nbsp;     or when receiving locally available IP multic=
ast or sending IP


   &=
nbsp;     multicast packets.


 


[Sri] Does RFC 67=
05, or RFC 6909 does not address this issue ? May be the CDN example is inc=
orrect. 


 


 


   PS2:&n=
bsp; Divergence from other evolutionary trends in network=



   &=
nbsp;     architectures such as distribution of content=
 delivery.


 


   &=
nbsp;     Centralized mobility management can become no=
n-optimal with a


   &=
nbsp;     flat network architecture.<=
/pre>

 


[Sri] How is this=
 making the case of DMM ? We want the MN to access content locally in the a=
ccess network and we want localized routing. We have that in the form of 67=
05 and 6909. The other approach is give localized IP addresses and have the=
 content locally accessed. There are simply two many considerations and poi=
nts may be valid, but when we bring all those assumptions. We are bringing =
these points in a less logical manner without stating the assumptions.


 


   PS3:&n=
bsp; Low scalability of centralized tunnel management and mobility


   &=
nbsp;     context maintenance


 


   &=
nbsp;     Setting up tunnels through a central anchor a=
nd maintaining


   &=
nbsp;     mobility context for each MN usually requires=
 more concentrated


   &=
nbsp;     resources in a centralized design, thus reduc=
ing scalability.


   &=
nbsp;     Distributing the tunnel maintenance function =
and the mobility


   &=
nbsp;     context maintenance function among different =
network entities


   &=
nbsp;     with proper signaling protocol design can inc=
rease scalability.


 


 


 


 


Chan (Ed.), et al.&=
nbsp;     Expires February 3, 2014   &nb=
sp;            [Page=
 9]


 


Internet-Draft =
;            &n=
bsp;    DMM-Reqs       &n=
bsp;            =
; August 2013


 


 


   PS4:&n=
bsp; Single point of failure and attack


 


   &=
nbsp;     Centralized anchoring designs may be more vul=
nerable to single


   &=
nbsp;     points of failures and attacks than a distrib=
uted system.  The


   &=
nbsp;     impact of a successful attack on a system wit=
h centralized


   &=
nbsp;     mobility management can be far greater as wel=
l.


 


   PS5:&n=
bsp; Unnecessarily reserving resources to provide mobility support


   &=
nbsp;     to nodes that do not need such support


 


   &=
nbsp;     IP mobility support is not always required, a=
nd not every


   &=
nbsp;     parameter of mobility context is always used.=
  For example,


   &=
nbsp;     some applications do not need a stable IP add=
ress during a


   &=
nbsp;     handover to maintain session continuity. =
; Sometimes, the entire


   &=
nbsp;     application session runs while the terminal d=
oes not change the


   &=
nbsp;     point of attachment.  Besides, some sess=
ions, e.g.  SIP-based


   &=
nbsp;     sessions, can handle mobility at the applicat=
ion layer and


   &=
nbsp;     hence do not need IP mobility support; it is =
then more


   &=
nbsp;     efficient to deactivate IP mobility support f=
or such sessions.


 


[Sri]  Mobility systems toda=
y do support the aspect of service for the subscriber, as "Simple IP&q=
uot;, or "Mobile IP". A PDSN can assign a local IP address and it=
 does not have to be the home address. Network does have this intelligence,=
 but what is missing is the client's ability to pick the correct type of IP=
 address, among different IP addresses. The network is also the missing the=
 aspect of marking those addresses with proper properties. The currently ac=
tive drafts in IETF are to address this issue. We should talk about these m=
issing semantics.<=
/pre>

draft-bhandari-dhc-class-based-prefix-05  


draft-korhonen-6man-prefix-properties-02<=
/span>


 


 


   PS6:&n=
bsp; (Related problem) Mobility signaling overhead with peer-to-peer


   &=
nbsp;     communication


 


   &=
nbsp;     Wasting resources when mobility signaling (e.=
g., maintenance of


   &=
nbsp;     the tunnel, keep alive signaling, etc.) is no=
t turned off for


   &=
nbsp;     peer-to-peer communication.  Peer-to-pee=
r communications have


   &=
nbsp;     particular traffic patterns that often do not=
 benefit from


   &=
nbsp;     mobility support from the network.  Thus=
, the associated


   &=
nbsp;     mobility support signaling (e.g., maintenance=
 of the tunnel,


   &=
nbsp;     keep alive signaling, etc.) wastes network re=
sources for no


   &=
nbsp;     application gain.  In such a case, it is=
 better to enable


   &=
nbsp;     mobility support selectively.


 


[Sri] How is PS6 different from P=
S5 ? We talk about application's ability to pick the address with or withou=
t mobility properties in PS5. So, the traffic patterns can be localized bas=
ed on the application requirements. But, even if we take the argument that =
mobility is not required, operator needs visibility into these flows and so=
 they can charge.  Again, we have 6705 and 6909 for adjusting to those=
 traffic patterns. So, this P without those considerations is incomplete.


 


 


   PS7:&n=
bsp; (Related problem) Deployment with multiple mobility solutions


 


   &=
nbsp;     There are already many variants and extension=
s of MIP.


    =
     Deployment of new mobility management solutio=
ns can be


   &=
nbsp;     challenging, and debugging difficult, when th=
ey must co-exist


   &=
nbsp;     with solutions already in the field.


 


 


 


   PS8:  Duplicate m=
ulticast traffic


 


      =
;   IP multicast distribution over architectures using IP mobilit=
y


      =
;   solutions (e.g.  RFC6224) may lead to convergence of dup=
licated


      =
;   multicast subscriptions towards the downstream tunnel entity<=
o:p>


      =
;   (e.g.  MAG in PMIPv6).  Concretely, when multicast =
subscription


      &nbs=
p;  for individual mobile nodes is coupled with mobility tunnels<=
o:p>


      =
;   (e.g.  PMIPv6 tunnel), duplicate multicast subscription(=
s) is


 


 


 


Chan (Ed.), et al.   =
;   Expires February 3, 2014      &=
nbsp;        [Page 10]=



 


Internet-Draft   &nb=
sp;            =
  DMM-Reqs          =
           August 2013


 


 


      =
;   prone to be received through different upstream paths.  =
This


      =
;   problem may also exist or be more severe in a distributed


      =
;   mobility environment.


 


[Sri] Is this for MN's from diffe=
rent LMA's attached to the same MAG ?


 


5.  Requiremen=
ts


 


   After =
comparing distributed mobility management against centralized


   deploy=
ment in Section 3, this section identifies the following<=
/pre>

   requir=
ements:


 


5.1.  Distribu=
ted processing


 


   REQ1:&=
nbsp; Distributed processing


 


   &=
nbsp;      IP mobility, network access and routing=
 solutions provided by


   &=
nbsp;      DMM MUST enable distributed processing =
for mobility management


   &=
nbsp;      so that traffic does not need to traver=
se centrally deployed


   &=
nbsp;      mobility anchors and thereby avoid non-=
optimal routes.


 


   &=
nbsp;      Motivation: This requirement is motivat=
ed by current trends in


   &=
nbsp;      network evolution: (a) it is cost- and =
resource-effective to


   &=
nbsp;      cache and distribute content by combini=
ng distributed mobility


   &=
nbsp;      anchors with caching systems (e.g., CDN=
); (b) the


   &=
nbsp;      significantly larger number of mobile n=
odes and flows call for


   &=
nbsp;      improved scalability; (c) single points=
 of failure are avoided


   &=
nbsp;      in a distributed system; (d) threats ag=
ainst centrally


   &=
nbsp;      deployed anchors, e.g., home agent and =
local mobility anchor,


   &=
nbsp;      are mitigated in a distributed system.<=
o:p>


 


   This r=
equirement addresses the problems PS1, PS2, PS3, and PS4<=
/pre>

   descri=
bed in Section 4.  (Existing route optimization is only a host-


   based =
solution.  On the other hand, localized routing with PMIPv6=



   addres=
ses only a part of the problem where both the MN and the CN are<=
/span>


   locate=
d in the PMIP domain and attached to a MAG, and is not

   applic=
able when the CN is outside the PMIP domain.)


 


[Sri] I'm still stuck on the CDN =
example driving this requirement.


 


 


5.2.  Transpar=
ency to Upper Layers when needed


 


   REQ2:&=
nbsp; Transparency to Upper Layers when needed


 


   &=
nbsp;      DMM solutions MUST provide transparent =
mobility support above


   &=
nbsp;      the IP layer when needed.  Such tr=
ansparency is needed, for


   &=
nbsp;      example, when, upon change of point of =
attachment to the


   &=
nbsp;      network, an application flow cannot cop=
e with a change in the


   &=
nbsp;      IP address.  However, it is not al=
ways necessary to maintain a


   &=
nbsp;      stable home IP address or prefix for ev=
ery application or at


   &=
nbsp;      all times for a mobile node.=



 


 


[Sri] Please refl=
ect the two key aspects of this requirement:


·        Network can assign I=
P addresses with different properties; It carries those properties


·        Applications have di=
fferent requirements and will pick the address with the correct property.


 


 


Chan (Ed.), et al.&=
nbsp;     Expires February 3, 2014   &nb=
sp;           [Page 11]


 


Internet-Draft =
;            &n=
bsp;    DMM-Reqs       &n=
bsp;            =
; August 2013


 


 


   &=
nbsp;      Motivation: The motivation of this requ=
irement is to enable


   &=
nbsp;      more efficient use of network resources=
 and more efficient


   &=
nbsp;      routing by not maintaining context at t=
he mobility anchor when


   &=
nbsp;      there is no such need.


   This r=
equirement addresses the problem PS5 as well as the related


   proble=
m PS6 stated in Section 4.


 


5.3.  IPv6 dep=
loyment


 


   REQ3:&=
nbsp; IPv6 deployment


 


   &=
nbsp;      DMM solutions SHOULD target IPv6 as the=
 primary deployment


   &=
nbsp;      environment and SHOULD NOT be tailored =
specifically to support


   &=
nbsp;      IPv4, in particular in situations where=
 private IPv4 addresses


   &=
nbsp;      and/or NATs are used.=



 


   &=
nbsp;      Motivation: This requirement conforms t=
o the general


   &=
nbsp;      orientation of IETF work.  DMM dep=
loyment is foreseen in mid-


   &=
nbsp;      to long-term horizon, when IPv6 is expe=
cted to be far more


   &=
nbsp;      common than today.

 


   This r=
equirement avoids the unnecessarily complexity in solving the


   proble=
ms in Section 4 for IPv4, which will not be able to use some of<=
/span>


   the IP=
v6-specific features.


 


5.4.  Existing=
 mobility protocols


 


   REQ4:&=
nbsp; Existing mobility protocols


 


   &=
nbsp;      A DMM solution SHOULD first consider re=
using and extending


   &=
nbsp;      IETF-standardized protocols before spec=
ifying new protocols.


 


   &=
nbsp;      Motivation: Reuse of existing IETF work=
 is more efficient and


   &=
nbsp;      less error-prone.

 


   This r=
equirement attempts to avoid the need of new protocols

   develo=
pment and therefore their potential problems of being time-


   consum=
ing and error-prone.


 


5.5.  Co-exist=
ence


 


   REQ5:&=
nbsp; Co-existence with deployed networks and hosts


 


   &=
nbsp;      The DMM solution MUST be able to co-exi=
st with existing


   &=
nbsp;      network deployments and end hosts. =
; For example, depending on


   &=
nbsp;      the environment in which DMM is deploye=
d, DMM solutions may


   &=
nbsp;      need to be compatible with other deploy=
ed mobility protocols


 


 


 


Chan (Ed.), et al.&=
nbsp;     Expires February 3, 2014   &nb=
sp;           [Page 12]


 


Internet-Draft =
;            &n=
bsp;    DMM-Reqs       &n=
bsp;            =
; August 2013


 


 


   &=
nbsp;      or may need to co-exist with a network =
or mobile hosts/routers


   &=
nbsp;      that do not support DMM protocols. =
; The mobile node may also


   &=
nbsp;      move between different access networks,=
 where some of them may


   &=
nbsp;      support neither DMM nor another mobilit=
y protocol.


   &=
nbsp;      Furthermore, a DMM solution SHOULD work=
 across different


   &=
nbsp;      networks, possibly operated as separate=
 administrative


   &=
nbsp;      domains, when allowed by the trust rela=
tionship between them.


 


   &=
nbsp;      Motivation: (a) to preserve backwards c=
ompatibility so that


   &=
nbsp;      existing networks and hosts are not aff=
ected and continue to


   &=
nbsp;      function as usual, and (b) enable inter=
-domain operation if


   &=
nbsp;      desired.


 


   This r=
equirement addresses the related problem PS7 described in=



   Sectio=
n 4.


 


5.6.  Security=
 considerations


 


   REQ6:&=
nbsp; Security considerations


 


   &=
nbsp;      A DMM solution MUST not introduce new s=
ecurity risks or


   &=
nbsp;      amplify existing security risks against=
 which the existing


   &=
nbsp;      security mechanisms/protocols cannot of=
fer sufficient


   &=
nbsp;      protection.


 


   &=
nbsp;      Motivation: Various attacks such as imp=
ersonation, denial of


   &=
nbsp;      service, man-in-the-middle attacks, and=
 so on, may be launched


   &=
nbsp;      in a DMM deployment.  For instance=
, an illegitimate node may


   &=
nbsp;      attempt to access a network providing D=
MM.  Another example is


   &=
nbsp;      that a malicious node can forge a numbe=
r of signaling messages


   &=
nbsp;      thus redirecting traffic from its legit=
imate path.


   &=
nbsp;      Consequently, the specific node is unde=
r a denial of service


   &=
nbsp;      attack, whereas other nodes do not rece=
ive their traffic.


   &=
nbsp;      Accordingly, security mechanisms/protoc=
ols providing access


   &=
nbsp;      control, integrity, authentication, aut=
horization,


   &=
nbsp;      confidentiality, etc. can be used to pr=
otect the DMM entities


   &=
nbsp;      as they are already used to protect aga=
inst existing networks


   &=
nbsp;      and existing mobility protocols defined=
 in IETF.  In addition,


   &=
nbsp;      end-to-end security measures between co=
mmunicating nodes may


   &=
nbsp;      already be used when deploying existing=
 mobility protocols


   &=
nbsp;      where the signaling messages travel ove=
r the Internet.  For


   &=
nbsp;      instance, EAP-based authentication can =
be used for network


   &=
nbsp;      access security, while IPsec can be use=
d for end-to-end


   &=
nbsp;      security.  When the existing secur=
ity mechanisms/protocols are


   &=
nbsp;      applied to protect the DMM entities, th=
e security risks that


   &=
nbsp;      may be introduced by DMM MUST be consid=
ered to be eliminated.


   &=
nbsp;      Else the security protection would be d=
egraded in the DMM


   &=
nbsp;      solution versus in existing mobility pr=
otocols.


 


 


 


Chan (Ed.), et al.&=
nbsp;     Expires February 3, 2014   &nb=
sp;           [Page 13]


 


Internet-Draft =
;            &n=
bsp;    DMM-Reqs       &n=
bsp;            =
; August 2013


 


 


   This r=
equirement prevents a DMM solution from introducing


   uncont=
rollable problems of potentially insecure mobility management


   protoc=
ols which make deployment infeasible because platforms

   confor=
ming to the protocols are at risk for data loss and numerous


   other =
dangers, including financial harm to the users.


 


5.7.  Multicas=
t


 


   REQ7:&=
nbsp; Multicast considerations


 


   &=
nbsp;      DMM SHOULD consider multicast early so =
that solutions can be


   &=
nbsp;      developed not only to provide IP mobili=
ty support when it is


   &=
nbsp;      needed, but also to avoid network ineff=
iciency issues in


   &=
nbsp;      multicast traffic delivery (such as dup=
licate multicast


   &=
nbsp;      subscriptions towards the downstream tu=
nnel entities).  The


   &=
nbsp;      multicast solutions should therefore av=
oid restricting the


   &=
nbsp;      management of all IP multicast traffic =
to a single host


   &=
nbsp;      through a dedicated (tunnel) interface =
on multicast-capable


   &=
nbsp;      access routers.


 


   &=
nbsp;      Motivation: Existing multicast deployme=
nt have been introduced


   &=
nbsp;      after completing the design of the refe=
rence mobility


   &=
nbsp;      protocol, then optimization and extensi=
ons have been followed


   &=
nbsp;      by "patching-up" procedure, t=
hus leading to network


   &=
nbsp;      inefficiency and non-optimal routing.&n=
bsp; The multicast solutions


   &=
nbsp;      should therefore be required to conside=
r efficiency nature in


   &=
nbsp;      multicast traffic delivery.<=
/span>


 


   This r=
equirement addresses the problems PS1 and PS8 described in


   Sectio=
n 4.


 


 


6.  Security C=
onsiderations


 


   Please=
 refer to the discussion under Security requirement in Section


   5.6.


 


 


7.  IANA Consi=
derations


 


   None


 


 


8.  Co-authors=
 and Contributors


 


   This p=
roblem statement document is a joint effort among the numerous


   partic=
ipants.  Each individual has made significant contributions to


   this w=
ork and have been listed as co-authors.


 


 


 


Chan (Ed.), et al.&=
nbsp;     Expires February 3, 2014   &nb=
sp;           [Page 14]


 


Internet-Draft =
;            &n=
bsp;    DMM-Reqs       &n=
bsp;            =
; August 2013


 


 


9.  References=



 


9.1.  Normativ=
e References


 


   [RFC21=
19]  Bradner, S., "Key words for use in RFCs to Indicate


   &=
nbsp;          Requirement Lev=
els", BCP 14, RFC 2119, March 1997.


 


9.2.  Informat=
ive References


 


   [I-D.y=
okota-dmm-scenario]


   &=
nbsp;          Yokota, H., Sei=
te, P., Demaria, E., and Z. Cao, "Use case


   &=
nbsp;          scenarios =
 for Distributed Mobility Management",


   &=
nbsp;          draft-yokota-dm=
m-scenario-00 (work in progress),


   &=
nbsp;          October 2010.


 


   [Paper=
-Distributed.Centralized.Mobility]


   &=
nbsp;          Bertin, P., Bon=
jour, S., and J-M. Bonnin, "A Distributed


   &=
nbsp;          or Centralized =
Mobility",  Proceedings of Global


   &=
nbsp;          Communications =
Conference  (GlobeCom), December 2009.


 


   [Paper=
-Distributed.Dynamic.Mobility]


   &=
nbsp;          Bertin, P., Bon=
jour, S., and J-M. Bonnin, "A Distributed


   &=
nbsp;          Dynamic Mobilit=
y Management Scheme  Designed for Flat IP


   &=
nbsp;          Architectures&q=
uot;,  Proceedings of 3rd International


   &=
nbsp;          Conference =
; on New Technologies, Mobility and Security


   &=
nbsp;          (NTMS), 2008.


 


   [Paper=
-Distributed.Mobility.MIP]


   &=
nbsp;          Chan, H., "=
;Distributed Mobility Management with Mobile


   &=
nbsp;          IP", =
 Proceedings of  IEEE International Communication

   &=
nbsp;          Conference (ICC=
)  Workshop on Telecommunications:  from


   &=
nbsp;          Research to Sta=
ndards, June 2012.


 


   [Paper=
-Distributed.Mobility.PMIP]


   &=
nbsp;          Chan, H., "=
;Proxy Mobile IP  with Distributed Mobility


   &=
nbsp;          Anchors",&=
nbsp; Proceedings of GlobeCom Workshop  on Seamless<=
/pre>

   &=
nbsp;          Wireless Mobili=
ty, December 2010.


 


   [Paper=
-Distributed.Mobility.Review]


   &=
nbsp;          Chan, H., Yokot=
a, H., Xie, J., Seite, P., and D. Liu,


   &=
nbsp;          "Distribut=
ed and Dynamic Mobility Management  in Mobile


   &=
nbsp;          Internet: Curre=
nt Approaches and Issues, Journal of


   &=
nbsp;          Communications,=
 vol. 6, no. 1, pp. 4-15, Feb 2011.",


   &=
nbsp;           Proceedin=
gs of GlobeCom Workshop  on Seamless Wireless


   &=
nbsp;          Mobility, Febru=
ary 2011.


 


   [Paper=
-Distributed.Mobility.SAE]


   &=
nbsp;          Fisher, M., And=
erson, F., Kopsel, A., Schafer, G., and M.


 


 


 


Chan (Ed.), et al.&=
nbsp;     Expires February 3, 2014   &nb=
sp;           [Page 15]


 


Internet-Draft =
;            &n=
bsp;    DMM-Reqs       &n=
bsp;            =
; August 2013


 


 


   &=
nbsp;          Schlager, "=
;A Distributed IP Mobility Approach for 3G SAE",

   &=
nbsp;           Proceedin=
gs of the 19th International Symposium  on


   &=
nbsp;          Personal, Indoo=
r and Mobile Radio Communications (PIMRC),


   &=
nbsp;          2008.


 


   [Paper=
-Locating.User]


   &=
nbsp;          Kirby, G., &quo=
t;Locating the User",  Communication


   &=
nbsp;          International, =
1995.


 


   [Paper=
-Migrating.Home.Agents]


   &=
nbsp;          Wakikawa, R., V=
aladon, G., and J. Murai, "Migrating Home


   &=
nbsp;          Agents  To=
wards Internet-scale Mobility Deployments",


   &=
nbsp;           Proceedin=
gs of the ACM 2nd CoNEXT Conference  on Future


    =
          Networking Tech=
nologies, December 2006.


 


   [Paper=
-Mobile.Data.Offloading]


   &=
nbsp;          Lee, K., Lee, J=
., Yi, Y., Rhee, I., and S. Chong, "Mobile


   &=
nbsp;          Data Offloading=
: How Much Can WiFi Deliver?",  SIGCOMM


   &=
nbsp;          2010, 2010.


 


   [RFC37=
53]  Manner, J. and M. Kojo, "Mobility Related Terminology",=



   &=
nbsp;          RFC 3753, June =
2004.


 


   [RFC52=
13]  Gundavelli, S., Leung, K., Devarapalli, V., Chowdhury, K.,


   &=
nbsp;          and B. Patil, &=
quot;Proxy Mobile IPv6", RFC 5213, August 2008.

 


   [RFC53=
80]  Soliman, H., Castelluccia, C., ElMalki, K., and L.


   &=
nbsp;          Bellier, "=
Hierarchical Mobile IPv6 (HMIPv6) Mobility


   &=
nbsp;          Management"=
;, RFC 5380, October 2008.


 


   [RFC59=
44]  Perkins, C., "IP Mobility Support for IPv4, Revised",


   &=
nbsp;          RFC 5944, Novem=
ber 2010.


 


   [RFC62=
75]  Perkins, C., Johnson, D., and J. Arkko, "Mobility Support


   &=
nbsp;          in IPv6", =
RFC 6275, July 2011.


 


   [RFC63=
01]  Zhu, Z., Wakikawa, R., and L. Zhang, "A Survey of Mobility


   &=
nbsp;          Support in the =
Internet", RFC 6301, July 2011.


 


   [TS.23=
.401]


   &=
nbsp;          3GPP, "Gen=
eral Packet Radio Service (GPRS) enhancements


   &=
nbsp;          for Evolved Uni=
versal Terrestrial Radio Access Network


   &=
nbsp;          (E-UTRAN) acces=
s", 3GPP TR 23.401 10.10.0, March 2013.


 


   [TS.29=
303]


   &=
nbsp;          3GPP, "Dom=
ain Name System Procedures; Stage 3", 3GPP


   &=
nbsp;          TR 23.303 11.2.=
0, September 2012.


 


 


 


 


Chan (Ed.), et al.&=
nbsp;     Expires February 3, 2014   &nb=
sp;           [Page 16]


 


Internet-Draft =
;            &n=
bsp;    DMM-Reqs       &n=
bsp;            &nbs=
p;August 2013


 


 


Authors' Addresses<=
o:p>


 


   H Anth=
ony Chan (editor)


   Huawei=
 Technologies (more co-authors on P. 17)


   5340 L=
egacy Dr. Building 3, Plano, TX 75024, USA


   Email:=
 h.a.chan@ieee.org


 


 


   Dapeng=
 Liu


   China =
Mobile


   Unit2,=
 28 Xuanwumenxi Ave, Xuanwu District,  Beijing 100053, China


   Email:=
 liudapeng@chinamobile.com=



 


 


   Pierri=
ck Seite


   Orange=



   4, rue=
 du Clos Courtel, BP 91226,  Cesson-Sevigne 35512, France


   Email:=
 pierrick.seite@orange.com=



 


 


   Hideto=
shi Yokota


   KDDI L=
ab


   2-1-15=
 Ohara, Fujimino, Saitama, 356-8502 Japan


   Email:=
 yokota@kddilabs.jp


 


 


   Jouni =
Korhonen


   Nokia =
Siemens Networks


   Email:=
 jouni.korhonen@nsn.com<=
/o:p>


   -=



   Charle=
s E. Perkins


   Huawei=
 Technologies


   Email:=
 charliep@computer.org


   -=



   Melia =
Telemaco


   Alcate=
l-Lucent Bell Labs


   Email:=
 telemaco.melia@alcate=
l-lucent.com


   -=



   Elena =
Demaria


   Teleco=
m Italia


   via G.=
 Reiss Romoli, 274, TORINO, 10148, Italy


   Email:=
 elena.demaria@telecomita=
lia.it


   -=



   Jong-H=
youk Lee


   RSM De=
partment, Telecom Bretagne


   Cesson=
-Sevigne, 35512, France


   Email:=
 jh.lee@telecom-bretagne.eu


   -=



 


 


 


Chan (Ed.), et al.&=
nbsp;     Expires February 3, 2014   &nb=
sp;           [Page 17]


 


Internet-Draft =
;            &n=
bsp;    DMM-Reqs       &n=
bsp;            =
; August 2013


 


 


   Kostas=
 Pentikousis


   Huawei=
 Technologies


   Carnot=
str. 4 10587 Berlin, Germany


   Email:=
 k.pentikousis@huawei.com


   -=



   Tricci=
 So


   ZTE


   Email:=
 tso@zteusa.com

   -=



   Carlos=
 J. Bernardos


   Univer=
sidad Carlos III de Madrid


   Av. Un=
iversidad, 30, Leganes, Madrid 28911, Spain


   Email:=
 cjbc@it.uc3m.es

   -=



   Peter =
McCann


   Huawei=
 Technologies


   Email:=
 PeterMcCann@huawei.com<=
/o:p>


   -=



   Seok J=
oo Koh


   Kyungp=
ook National University, Korea


   Email:=
 sjkoh@knu.ac.kr

   -=



   Wen Lu=
o


   ZTE


   No.68,=
 Zijinhua RD,Yuhuatai District, Nanjing, Jiangsu 210012, China


   Email:=
 luo.wen@zte.com.cn


   -=



   Sri Gu=
ndavelli


   sgundave@cisco.com

   -=



   Marco =
Liebsch


   NEC La=
boratories Europe


   Email:=
 liebsch@neclab.eu


   -=



   Carl W=
illiams


   MCSR L=
abs


   Email:=
 carlw@mcsr-labs.org<=
/span>


   -=



   Seil J=
eon


   Instit=
uto de Telecomunicacoes, Aveiro


   Email:=
 seiljeon@av.it.pt


   -=



   Sergio=
 Figueiredo


   Univer=
sidade de Aveiro


   Email:=
 sfigueiredo@av.it.pt


   -=



   Stig V=
enaas


   Email:=
 stig@venaas.com

 


 


 


Chan (Ed.), et al.&=
nbsp;     Expires February 3, 2014   &nb=
sp;           [Page 18]


 


Internet-Draft =
;             &=
nbsp;   DMM-Reqs       &n=
bsp;            =
; August 2013


 


 


   -=



   Luis M=
iguel Contreras Murillo


   Email:=
 lmcm@tid.es


   -=



   Juan C=
arlos Zuniga


   Email:=
 JuanCarlos.Zuniga@In=
terDigital.com


   -=



   Alexan=
dru Petrescu


   Email:=
 alexandru.petrescu@gmail.c=
om


   -=



   Georgi=
os Karagiannis


   Email:=
 g.karagiannis@utwente.nl


   -=



   Julien=
 Laganier


   jlaganier@juniper.net


   -=



   Wassim=
 Michel Haddad


   Wassam.Haddad@ericsson.com=



   -=



   Dirk v=
on Hugo


   Dirk.von-Hugo@telekom.de


   -=



   Ahmad =
Muhanna


   amuhanna@awardsolutions.com


   -=



   Byoung=
-Jo Kim


   ATT La=
bs


   macsbug@research.att.com


   -=



   Hassan=
 Aliahmad


   Orange=



   hassan.aliahmad@orange.com=



   -=



 


 


 


 


 


 


 


 


 


 


 


 


 


 


 


 


 


 


Chan (Ed.), et al.&=
nbsp;     Expires February 3, 2014   &nb=
sp;           [Page 19]

 

--_000_6E31144C030982429702B11D6746B98C370CBC16szxeml557mbxchi_-- From h.anthony.chan@huawei.com Wed Sep 25 22:50:21 2013 Return-Path: X-Original-To: dmm@ietfa.amsl.com Delivered-To: dmm@ietfa.amsl.com Received: from localhost (localhost [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id E66EB21F9991 for ; Wed, 25 Sep 2013 22:50:20 -0700 (PDT) X-Virus-Scanned: amavisd-new at amsl.com X-Spam-Flag: NO X-Spam-Score: -6.283 X-Spam-Level: X-Spam-Status: No, score=-6.283 tagged_above=-999 required=5 tests=[BAYES_00=-2.599, HTML_MESSAGE=0.001, RCVD_IN_DNSWL_MED=-4, SARE_MILLIONSOF=0.315] Received: from mail.ietf.org ([12.22.58.30]) by localhost (ietfa.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id EYEsIUBLNXIf for ; Wed, 25 Sep 2013 22:50:12 -0700 (PDT) Received: from lhrrgout.huawei.com (lhrrgout.huawei.com [194.213.3.17]) by ietfa.amsl.com (Postfix) with ESMTP id E40D111E8140 for ; Wed, 25 Sep 2013 22:50:10 -0700 (PDT) Received: from 172.18.7.190 (EHLO lhreml203-edg.china.huawei.com) ([172.18.7.190]) by lhrrg01-dlp.huawei.com (MOS 4.3.7-GA FastPath queued) with ESMTP id AYH11233; Thu, 26 Sep 2013 05:50:07 +0000 (GMT) Received: from LHREML405-HUB.china.huawei.com (10.201.5.242) by lhreml203-edg.huawei.com (172.18.7.221) with Microsoft SMTP Server (TLS) id 14.3.146.0; Thu, 26 Sep 2013 06:49:04 +0100 Received: from SZXEML411-HUB.china.huawei.com (10.82.67.138) by lhreml405-hub.china.huawei.com (10.201.5.242) with Microsoft SMTP Server (TLS) id 14.3.146.0; Thu, 26 Sep 2013 06:49:57 +0100 Received: from szxeml557-mbx.china.huawei.com ([169.254.5.217]) by szxeml411-hub.china.huawei.com ([::1]) with mapi id 14.03.0146.000; Thu, 26 Sep 2013 13:49:51 +0800 From: h chan To: "Sri Gundavelli (sgundave)" , "dmm@ietf.org" Thread-Topic: [DMM] I-D Action: draft-ietf-dmm-requirements-07.txt Thread-Index: AQHOogjoa3/L9muNGkS5wozrKtEQEpnXRZkQgABo14CAAAG/AA== Date: Thu, 26 Sep 2013 05:49:51 +0000 Message-ID: <6E31144C030982429702B11D6746B98C370CBC1C@szxeml557-mbx.china.huawei.com> References: <1377486056.79906.YahooMailNeo@web163805.mail.gq1.yahoo.com> <24C0F3E22276D9438D6F366EB89FAEA8116A4D4F@xmb-aln-x03.cisco.com> Accept-Language: en-US, zh-CN Content-Language: en-US X-MS-Has-Attach: X-MS-TNEF-Correlator: x-originating-ip: [10.46.81.5] Content-Type: multipart/alternative; boundary="_000_6E31144C030982429702B11D6746B98C370CBC1Cszxeml557mbxchi_" MIME-Version: 1.0 X-CFilter-Loop: Reflected Subject: Re: [DMM] I-D Action: draft-ietf-dmm-requirements-07.txt X-BeenThere: dmm@ietf.org X-Mailman-Version: 2.1.12 Precedence: list List-Id: Distributed Mobility Management Working Group List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Thu, 26 Sep 2013 05:50:21 -0000 --_000_6E31144C030982429702B11D6746B98C370CBC1Cszxeml557mbxchi_ Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: quoted-printable Sri, Thanks for the comments. We are replying under different sections in separa= te emails. ----------------------------------------------- Regarding your comments on Section 1, this section has been revised and reo= rganized in version 08 to avoid misunderstanding. Also our responses in-lin= e. Please check. 1. Introduction In the past decade a fair number of mobility protocols have been standardized [RFC6275] [RFC5944] [RFC5380] [RFC6301] [RFC5213]. Although the protocols differ in terms of functions and associated message formats, they all employ a mobility anchor to allow a mobile node to remain reachable after it has moved to a different network. The anchor point, among other tasks, ensures connectivity by forwarding packets destined to, or sent from, the mobile node. It is a centrally deployed mobility anchor in the sense that the deployed architectures today have a small number of these anchors and the traffic of millions of mobile nodes in an operator network are typically managed by the same anchor. Distributed mobility management (DMM) is an alternative to the above centralized deployment. The background behind the interests to study DMM are primarily in the following. (1) Mobile users are, more than ever, consuming Internet content; such traffic imposes new requirements on mobile core networks for data traffic delivery. The presence of content providers closer to Internet Service Providers (ISP) network requires taking into account local Content Delivery Networks (CDNs) while providing mobility services. Moreover, when the traffic demand exceeds available capacity, service providers need to implement new strategies such as selective traffic offload (e.g. [RFC6909], 3GPP work items LIPA/SIPTO [TS.23.401]) through alternative access networks (e.g. WLAN) [Paper- Mobile.Data.Offloading]. A gateway selection mechanism also takes the user proximity into account within EPC [TS.29303]. These mechanisms were not pursued in the past owing to charging and billing reasons. Assigning a gateway anchor node from a visited network in roaming scenario has until recently been done and are limited to voice services only. Charging and billing require solutions beyond the mobility protocol. Both traffic offloading and CDN mechanisms could benefit from the development of mobile architectures with fewer levels of routing hierarchy introduced into the data path by the mobility management system. This trend towards so-called "flat networks" works best for direct communications among peers in the same geographical area. Distributed mobility management in a truly flat mobile architecture would anchor the traffic closer to the point of attachment of the user. (2) Today's mobile networks present service providers with new challenges. Mobility patterns indicate that mobile nodes often remain attached to the same point of attachment for considerable periods of time [Paper-Locating.User]. Specific IP mobility management support is not required for applications that launch and complete their sessions while the mobile node is connected to the same point of attachment. However, currently, IP mobility support is designed for always-on operation, maintaining all parameters of the context for each mobile subscriber for as long as they are connected to the network. This can result in a waste of resources and unnecessary costs for the service provider. Infrequent node mobility coupled with application intelligence suggest that mobility support could be provided selectively, thus reducing the amount of context maintained in the network. In addition, considerations in the study of DMM are in the following. (1) To optimize handovers from the perspective of mobile nodes, the base protocols have been extended to efficiently handle packet forwarding between the previous and new points of attachment. These extensions are necessary when applications have stringent requirements in terms of delay. Notions of localization and distribution of local agents have been introduced to reduce signaling overhead at the centralized routing anchor point [Paper-Distributed.Centralized.Mobility]. Unfortunately, such protocols have not been deployed today. (2) Most existing mobility protocols have not been designed for multiple-interface hosts which are capable to use multiple interfaces simultaneously. Retrofitting the required functionality can result in an unnecessary increase in the protocol complexity. (3) IP multicast support, including optimizations, have been introduced as an effective transport method for multimedia data delivery, but by "patching-up" procedure after completing the design of reference mobility protocol, leading to network inefficiency and non-optimal routing. The distributed mobility management (DMM) charter addresses two complementary aspects of mobility management procedures: the distribution of mobility anchors in the data-plane towards a more flat network and the selective activation/deactivation of mobility protocol support as an enabler to distributed mobility management. The former aims at positioning mobility anchors (e.g., HA, LMA) closer to the user; ideally, mobility agents could be collocated with the first-hop router. The latter, facilitated by the distribution of mobility anchors, identifies when mobility support must be activated and when sessions do not require mobility management support -- thus reducing the amount of state information that must be maintained in various mobility agents of the mobile network. It can then avoid the unnecessary establishment of mechanisms to forward traffic from an old to a new mobility anchor. This document compares distributed mobility management with centralized mobility management in Section 3. The problems that can be addressed with DMM are summarized in Section 4. The mandatory requirements as well as the optional requirements are given in Section 5. Finally, security considerations are discussed in Section 6. The problem statement and the use cases [I-D.yokota-dmm-scenario] can be found in [Paper-Distributed.Mobility.Review]. The following have contributed to the revisions in draft 08: Pierrick, Dape= ng, and Jouni From: dmm-bounces@ietf.org [mailto:dmm-bounces= @ietf.org] On Behalf Of Sri Gundavelli (sgundave) Sent: Sunday, August 25, 2013 10:04 PM To: dmm@ietf.org Subject: Re: [DMM] I-D Action: draft-ietf-dmm-requirements-07.txt Please see inline for some comments. Regards Sri .... 1. Introduction In the past decade a fair number of mobility protocols have been standardized [RFC6275] [RFC5944] [RFC5380] [RFC6301] [RFC5213]. Although the protocols differ in terms of functions and associated message formats, we can identify a few key common features: o a centralized mobility anchor providing global reachability and an always-on experience to the user; o extensions to the base protocols to optimize handover performance while users roam across wireless cells; and o extensions to enable the use of heterogeneous wireless interfaces for multi-mode terminals (e.g. smartphones). [Sri] Currently defined mobility protocols support handovers and multiple a= ccess technologies ? Not sure, how these two "common features" make the ca= se for DMM If the point is about centralized anchor's, you don't need the = other two points. Accept and revised in version 08 The presence of the centralized mobility anchor allows a mobile node to remain reachable after it has moved to a different network. The anchor point, among other tasks, ensures connectivity by forwarding packets destined to, or sent from, the mobile node. In practice, most of the deployed architectures today have a small number of centralized anchors managing the traffic of millions of mobile nodes. Compared with a distributed approach, a centralized approach is likely to have several issues or limitations affecting performance and scalability, which require costly network engineering to resolve. [Sri] All 3G/4G systems are based on this model and they are running fine. = Again, we don't need to argue against centralized model to justify distribu= ted model. Please simply state distributed model can be useful in certain d= eployments and hence the motivation for this work. Accept and revised in version 08 To optimize handovers from the perspective of mobile nodes, the base protocols have been extended to efficiently handle packet forwarding between the previous and new points of attachment. These extensions are necessary when applications have stringent requirements in terms of delay. Notions of localization and distribution of local agents have been introduced to reduce signaling overhead at the centralized routing anchor point [Paper-Distributed.Centralized.Mobility]. Unfortunately, today we witness difficulties in getting such protocols deployed, resulting in sub-optimal choices for the network Operators. [Sri] I assume this is about hierarchical models / Chaining ? What are thes= e "difficulties in getting such protocols deployed" ? Accept and revised in version 08 Moreover, the availability of multiple-interface host and the possibility of using several network interfaces simultaneously have motivated the development of even more protocol extensions to add more capabilities to the mobility management protocol. In the end, deployment is further complicated with the multitude of extensions. [Sri] Not sure I follow. Mobile IP protocols are in general access agnostic= . Not sure, what are the extensions that we have added on access-basis. Re-organize Section 1 in version 08 to clarify. As an effective transport method for multimedia data delivery, IP multicast support, including optimizations, have been introduced but by "patching-up" procedure after completing the design of reference mobility protocol, leading to network inefficiency and non-optimal routing. [Sri] Multicast related extensions have nothing to do with multi-access sup= port/host's capability to support multiple interfaces. Can you clarify ? Th= is text is not clear. Re-organize Section 1 in version 08 to clarify. Mobile users are, more than ever, consuming Internet content; such traffic imposes new requirements on mobile core networks for data traffic delivery. The presence of content providers closer to Internet Service Providers (ISP) network requires taking into account local Content Delivery Networks (CDNs) while providing mobility services. Moreover, when the traffic demand exceeds available capacity, service providers need to implement new strategies such as selective traffic offload (e.g. 3GPP work items LIPA/SIPTO [TS.23.401]) through alternative access networks (e.g. WLAN) [Paper- Mobile.Data.Offloading]. [Sri] Please add reference to IETF SIPTO doc, RFC6909, before 23.401 :) Accept and revised in version 08 A gateway selection mechanism also takes the user proximity into account within EPC [TS.29303]. These mechanisms were not pursued in the past owing to charging and billing reasons. Assigning a gateway anchor node from a visited network in roaming scenario has until recently been done and are limited to voice services only. Charging and billing require solutions beyond the mobility protocol. Both traffic offloading and CDN mechanisms could benefit from the development of mobile architectures with fewer levels of routing hierarchy introduced into the data path by the mobility management system. This trend towards so-called "flat networks" works best for direct communications among peers in the same geographical area. Distributed mobility management in a truly flat mobile architecture would anchor the traffic closer to the point of attachment of the user. Today's mobile networks present service providers with new challenges. Mobility patterns indicate that mobile nodes often remain attached to the same point of attachment for considerable periods of time [Paper-Locating.User]. Specific IP mobility management support is not required for applications that launch and complete their sessions while the mobile node is connected to the same point of attachment. However, currently, IP mobility support is designed for always-on operation, maintaining all parameters of the context for each mobile subscriber for as long as they are connected to the network. This can result in a waste of resources and unnecessary= costs for the service provider. [Sri] Is the intent of this is text is about routing based approaches ? If a mobile is attached to the network, it does have some state at the anch= or. Also, if we consider the home link in mobility models, there is no stat= e for the mobile node when it is at home. In case of DMM, or with the curre= nt models, if the gateway selection is based on the MN's location and when = there is no node mobility, there should not be any state at the anchor. CDM= A We are probably talking about the same thing. Revise in 08 to clarify. Infrequent node mobility coupled with application intelligence suggest that mobility support could be provided selectively, thus reducing the amount of context maintained in the network. The distributed mobility management (DMM) charter addresses two complementary aspects of mobility management procedures: the distribution of mobility anchors towards a more flat network and the dynamic activation/deactivation of mobility protocol support as an enabler to distributed mobility management. [Sri] Not sure, I follow this point on Dynamic activation/de-activation. Ca= n you clarify. Re-organize Section 1 in version 08 to clarify. The former aims at positioning mobility anchors (e.g., HA, LMA) closer to the user; ideally, mobility agents could be collocated with the first-hop Chan (Ed.), et al. Expires February 3, 2014 [Page 5] Internet-Draft DMM-Reqs August 2013 router. The latter, facilitated by the distribution of mobility anchors, aims at identifying when mobility support must be activated and identifying sessions that do not require mobility management support -- thus reducing the amount of state information that must be maintained in various mobility agents of the mobile network. The key idea is that dynamic mobility management relaxes some of the constraints of previously-standardized mobility management solutions and, by doing so, it can avoid the unnecessary establishment of mechanisms to forward traffic from an old to a new mobility anchor. [Sri] The DMM model should not exclude the case of centralized anchor and d= istributed data plane. This is sub-case of DMM. Revise Section 1 in version 08 to clarify distribution for "data-plane". This document compares distributed mobility management with centralized mobility management in Section 3. The problems that can be addressed with DMM are summarized in Section 4. The mandatory requirements as well as the optional requirements are given in Section 5. Finally, security considerations are discussed in Section 6. The problem statement and the use cases [I-D.yokota-dmm-scenario] can be found in [Paper-Distributed.Mobility.Review]. --_000_6E31144C030982429702B11D6746B98C370CBC1Cszxeml557mbxchi_ Content-Type: text/html; charset="us-ascii" Content-Transfer-Encoding: quoted-printable

 <= /p>

Sri,

 <= /p>

Thanks for the comments. = We are replying under different sections in separate emails.

-------------------------= ----------------------

Regarding your comments o= n Section 1, this section has been revised and reorganized in version 08 to= avoid misunderstanding. Also our responses in-line. Please check.

 <= /p>

1.  Introduction

 

   In the past decade a fair number of mobility = protocols have been

   standardized [RFC6275] [RFC5944] [RFC5380] [R= FC6301] [RFC5213].

   Although the protocols differ in terms of fun= ctions and associated

   message formats, they all employ a mobility a= nchor to allow a mobile

   node to remain reachable after it has moved t= o a different network.

   The anchor point, among other tasks, ensures = connectivity by

   forwarding packets destined to, or sent from,= the mobile node.  It is

   a centrally deployed mobility anchor in the s= ense that the deployed

   architectures today have a small number of th= ese anchors and the

   traffic of millions of mobile nodes in an ope= rator network are

   typically managed by the same anchor.

 

   Distributed mobility management (DMM) is an a= lternative to the above

   centralized deployment.  The background = behind the interests to study

   DMM are primarily in the following.

 

   (1)  Mobile users are, more than ever, c= onsuming Internet content;

        such traffic im= poses new requirements on mobile core networks

        for data traffi= c delivery.  The presence of content providers

        closer to Inter= net Service Providers (ISP) network requires

        taking into acc= ount local Content Delivery Networks (CDNs) while

        providing mobil= ity services.  Moreover, when the traffic demand

        exceeds availab= le capacity, service providers need to implement

        new strategies = such as selective traffic offload (e.g.

        [RFC6909], 3GPP= work items LIPA/SIPTO [TS.23.401]) through

        alternative acc= ess networks (e.g.  WLAN) [Paper-

        Mobile.Data.Off= loading].  A gateway selection mechanism also

        takes the user = proximity into account within EPC [TS.29303].

        These mechanism= s were not pursued in the past owing to charging

        and billing rea= sons.  Assigning a gateway anchor node from a

        visited network= in roaming scenario has until recently been done

        and are limited= to voice services only.  Charging and billing

        require solutio= ns beyond the mobility protocol.

 

        Both traffic of= floading and CDN mechanisms could benefit from

        the development= of mobile architectures with fewer levels of

        routing hierarc= hy introduced into the data path by the mobility

        management syst= em.  This trend towards so-called "flat networks"=

        works best for = direct communications among peers in the same

        geographical ar= ea.  Distributed mobility management in a truly

        flat mobile arc= hitecture would anchor the traffic closer to the

        point of attach= ment of the user.

 

   (2)  Today's mobile networks present ser= vice providers with new

        challenges.&nbs= p; Mobility patterns indicate that mobile nodes often

        remain attached= to the same point of attachment for considerable

        periods of time= [Paper-Locating.User].  Specific IP mobility

        management supp= ort is not required for applications that launch

        and complete th= eir sessions while the mobile node is connected

        to the same poi= nt of attachment.  However, currently, IP

        mobility suppor= t is designed for always-on operation,

        maintaining all= parameters of the context for each mobile

        subscriber for = as long as they are connected to the network.

        This can result= in a waste of resources and unnecessary costs

        for the service= provider.  Infrequent node mobility coupled with

        application int= elligence suggest that mobility support could be

        provided select= ively, thus reducing the amount of context

        maintained in t= he network.

 

   In addition, considerations in the study of D= MM are in the following.

 

   (1)  To optimize handovers from the pers= pective of mobile nodes, the

        base protocols = have been extended to efficiently handle packet

        forwarding betw= een the previous and new points of attachment.

        These extension= s are necessary when applications have stringent

        requirements in= terms of delay.  Notions of localization and

        distribution of= local agents have been introduced to reduce

        signaling overh= ead at the centralized routing anchor point

        [Paper-Distribu= ted.Centralized.Mobility].  Unfortunately, such

        protocols have = not been deployed today.

 

   (2)  Most existing mobility protocols ha= ve not been designed for

        multiple-interf= ace hosts which are capable to use multiple

        interfaces simu= ltaneously.  Retrofitting the required

        functionality c= an result in an unnecessary increase in the

        protocol comple= xity.

 

   (3)  IP multicast support, including opt= imizations, have been

        introduced as a= n effective transport method for multimedia data

        delivery, but b= y "patching-up" procedure after completing the<= /p>

        design of refer= ence mobility protocol, leading to network

        inefficiency an= d non-optimal routing.

 

   The distributed mobility management (DMM) cha= rter addresses two

   complementary aspects of mobility management = procedures: the

   distribution of mobility anchors in the data-= plane towards a more

   flat network and the selective activation/dea= ctivation of mobility

   protocol support as an enabler to distributed= mobility management.

   The former aims at positioning mobility ancho= rs (e.g., HA, LMA)

   closer to the user; ideally, mobility agents = could be collocated with

   the first-hop router.  The latter, facil= itated by the distribution of

   mobility anchors, identifies when mobility su= pport must be activated

   and when sessions do not require mobility man= agement support -- thus

   reducing the amount of state information that= must be maintained in

   various mobility agents of the mobile network= .  It can then avoid the

   unnecessary establishment of mechanisms to fo= rward traffic from an

   old to a new mobility anchor.

 

   This document compares distributed mobility m= anagement with

   centralized mobility management in Section 3.=   The problems that can

   be addressed with DMM are summarized in Secti= on 4.  The mandatory

   requirements as well as the optional requirem= ents are given in

   Section 5.  Finally, security considerat= ions are discussed in Section

   6.

 

   The problem statement and the use cases [I-D.= yokota-dmm-scenario] can

   be found in [Paper-Distributed.Mobility.Revie= w].

 <= /p>

The following have contri= buted to the revisions in draft 08: Pierrick, Dapeng, and Jouni<= /span>

 <= /p>

 <= /p>

From: dmm-bounces@ietf.org [mailto:dmm-bounces@ietf.org] On Behalf Of Sri Gundavelli (sgundave)
Sent: Sunday, August 25, 2013 10:04 PM
To: dmm@ietf.org
Subject: Re: [DMM] I-D Action: draft-ietf-dmm-requirements-07.txt

 

Please see inline for some comments.

 

Regards

Sri

….
 
1.  Introducti=
on
 
   In the=
 past decade a fair number of mobility protocols have been
   standa=
rdized [RFC6275] [RFC5944] [RFC5380] [RFC6301] [RFC5213].=

   Althou=
gh the protocols differ in terms of functions and associated
   messag=
e formats, we can identify a few key common features:

 


   o =
; a centralized mobility anchor providing global reachability and an


   &=
nbsp;  always-on experience to the user;


 


   o =
; extensions to the base protocols to optimize handover performance


   &=
nbsp;  while users roam across wireless cells; and

 


   o =
; extensions to enable the use of heterogeneous wireless interfaces


   &=
nbsp;  for multi-mode terminals (e.g. smartphones).<=
/pre>

 


[Sri] Currently defined mobility =
protocols support handovers and multiple access technologies  ? Not su=
re, how these two "common features" make the case for DMM  <=
/span>If the point is about centralized ancho=
r's, you don't need the other two points. 


 =



Accept and revised i=
n version 08


 


   The pr=
esence of the centralized mobility anchor allows a mobile node


   to rem=
ain reachable after it has moved to a different network.  The


   anchor=
 point, among other tasks, ensures connectivity by forwarding


   packet=
s destined to, or sent from, the mobile node.  In practice,=



   most o=
f the deployed architectures today have a small number of=



   centra=
lized anchors managing the traffic of millions of mobile nodes.<=
/span>


   Compar=
ed with a distributed approach, a centralized approach is=



   likely=
 to have several issues or limitations affecting performance


   and sc=
alability, which require costly network engineering to resolve.<=
/span>


 


[Sri] All 3G/4G s=
ystems are based on this model and they are running fine. Again, we donR=
17;t need to argue against centralized model to justify distributed model. =
Please simply state distributed model can be useful in certain deployments =
and hence the motivation for this work.




 


Accept and revised i=
n version 08


   To opt=
imize handovers from the perspective of mobile nodes, the base


   protoc=
ols have been extended to efficiently handle packet forwarding


   betwee=
n the previous and new points of attachment.  These extensions


   are ne=
cessary when applications have stringent requirements in terms


   of del=
ay.  Notions of localization and distribution of local agents


   have b=
een introduced to reduce signaling overhead at the centralized


   routin=
g anchor point [Paper-Distributed.Centralized.Mobility].<=
/pre>

   Unfort=
unately, today we witness difficulties in getting such

   protoc=
ols deployed, resulting in sub-optimal choices for the network


   Operat=
ors.


[Sri] I assume th=
is is about hierarchical models / Chaining ? What are these "difficult=
ies in getting such protocols deployed" ?   


Accept and revised i=
n version 08


 


   M=
oreover, the availability of multiple-interface host and the


   possib=
ility of using several network interfaces simultaneously have


   motiva=
ted the development of even more protocol extensions to add


   more c=
apabilities to the mobility management protocol.  In the end,


   deploy=
ment is further complicated with the multitude of extensions.


[Sri] Not sure I follow. Mobile I=
P protocols are in general access agnostic. Not sure, what are the extensio=
ns that we have added on access-basis.=



 


Re-organize Section =
1 in version 08 to clarify. 


 


   As an =
effective transport method for multimedia data delivery, IP


   multic=
ast support, including optimizations, have been introduced but


   by &qu=
ot;patching-up" procedure after completing the design of reference


   mobili=
ty protocol, leading to network inefficiency and non-optimal


   routin=
g.


 


[Sri] Multicast r=
elated extensions have nothing to do with multi-access support/host's capab=
ility to support multiple interfaces. Can you clarify ? This text is not cl=
ear.


Re-organize Section =
1 in version 08 to clarify. 


 


 


   Mobile=
 users are, more than ever, consuming Internet content; such


   traffi=
c imposes new requirements on mobile core networks for data


   traffi=
c delivery.  The presence of content providers closer to


   Intern=
et Service Providers (ISP) network requires taking into account<=
/span>


   local =
Content Delivery Networks (CDNs) while providing mobility=



   servic=
es.  Moreover, when the traffic demand exceeds available


   capaci=
ty, service providers need to implement new strategies such as


   select=
ive traffic offload (e.g. 3GPP work items LIPA/SIPTO

   [TS.23=
.401]) through alternative access networks (e.g.  WLAN) [Paper-


   Mobile=
.Data.Offloading].  


[Sri] Please add =
reference to IETF SIPTO doc, RFC6909, before 23.401 :)


 


Accept and revised i=
n version 08


 =



A gateway selection=
 mechanism also takes


   the us=
er proximity into account within EPC [TS.29303].  These


   mechan=
isms were not pursued in the past owing to charging and billing<=
/span>


   reason=
s.   Assigning a gateway anchor node from a visited network in


   roamin=
g scenario has until recently been done and are limited to


   voice =
services only.  Charging and billing require solutions beyond


   the mo=
bility protocol.


 


   Both t=
raffic offloading and CDN mechanisms could benefit from the


   develo=
pment of mobile architectures with fewer levels of routing


   hierar=
chy introduced into the data path by the mobility management


   system=
.  This trend towards so-called "flat networks" works best f=
or


   direct=
 communications among peers in the same geographical area.


   Distri=
buted mobility management in a truly flat mobile architecture


   would =
anchor the traffic closer to the point of attachment of the


   user.<=
o:p>


 


   Today'=
s mobile networks present service providers with new

   challe=
nges.  Mobility patterns indicate that mobile nodes often


   remain=
 attached to the same point of attachment for considerable


   period=
s of time [Paper-Locating.User].  Specific IP mobility


   manage=
ment support is not required for applications that launch and


   comple=
te their sessions while the mobile node is connected to the


   same p=
oint of attachment. 


 However, currently=
, IP mobility support is


   design=
ed for always-on operation, maintaining all parameters of the


   contex=
t for each mobile subscriber for as long as they are connected


   to the=
 network.  This can result in a waste of resources and unnecessar=
y costs for the service provider. 


[Sri]  Is th=
e intent of this is text is about routing based approaches ?


If a mobile is at=
tached to the network, it does have some state at the anchor. Also, if we c=
onsider the home link in mobility models, there is no state for the mobile =
node when it is at home. In case of DMM, or with the current models, if the=
 gateway selection is based on the MN's location and when there is no node =
mobility, there should not be any state at the anchor. CDMA


 


We are probably talki=
ng about the same thing. Revise in 08 to clarify.


 


 Infrequent node mo=
bility


   couple=
d with application intelligence suggest that mobility support


   could =
be provided selectively, thus reducing the amount of context


   mainta=
ined in the network.


 


   The di=
stributed mobility management (DMM) charter addresses two=



   comple=
mentary aspects of mobility management procedures: the

   distri=
bution of mobility anchors towards a more flat network and the


   dynami=
c activation/deactivation of mobility protocol support as an


   enable=
r to distributed mobility management.  


[Sri] Not sure, I=
 follow this point on Dynamic activation/de-activation. Can you clarify.


 


Re-organize Section =
1 in version 08 to clarify. 


 =



The former aims at<=
o:p>


   positi=
oning mobility anchors (e.g., HA, LMA) closer to the user;


   ideall=
y, mobility agents could be collocated with the first-hop=



 


 


 


Chan (Ed.), et al.&=
nbsp;     Expires February 3, 2014   &nb=
sp;            [Page=
 5]


 


Internet-Draft =
;            &n=
bsp;    DMM-Reqs       &n=
bsp;            =
; August 2013


 


 


   router=
.  The latter, facilitated by the distribution of mobility<=
/span>


   anchor=
s, aims at identifying when mobility support must be activated


   and id=
entifying sessions that do not require mobility management


   suppor=
t -- thus reducing the amount of state information that must be<=
/span>


   mainta=
ined in various mobility agents of the mobile network.  The key


   idea i=
s that dynamic mobility management relaxes some of the

   constr=
aints of previously-standardized mobility management solutions


   and, b=
y doing so, it can avoid the unnecessary establishment of=



   mechan=
isms to forward traffic from an old to a new mobility anchor.


 


[Sri] The DMM mod=
el should not exclude the case of centralized anchor and distributed data p=
lane. This is sub-case of DMM. 


 


Revise Section 1 in =
version 08 to clarify distribution for “data-plane”. 


 =



   This d=
ocument compares distributed mobility management with

   centra=
lized mobility management in Section 3.  The problems that can


   be add=
ressed with DMM are summarized in Section 4.  The mandatory=



   requir=
ements as well as the optional requirements are given in<=
/pre>

   Sectio=
n 5.  Finally, security considerations are discussed in Section


   6.


 


   The pr=
oblem statement and the use cases [I-D.yokota-dmm-scenario] can<=
/span>


   be fou=
nd in [Paper-Distributed.Mobility.Review].


 


 

 

--_000_6E31144C030982429702B11D6746B98C370CBC1Cszxeml557mbxchi_-- From h.anthony.chan@huawei.com Wed Sep 25 22:55:43 2013 Return-Path: X-Original-To: dmm@ietfa.amsl.com Delivered-To: dmm@ietfa.amsl.com Received: from localhost (localhost [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id 2D70C21F9991 for ; Wed, 25 Sep 2013 22:55:43 -0700 (PDT) X-Virus-Scanned: amavisd-new at amsl.com X-Spam-Flag: NO X-Spam-Score: -6.283 X-Spam-Level: X-Spam-Status: No, score=-6.283 tagged_above=-999 required=5 tests=[BAYES_00=-2.599, HTML_MESSAGE=0.001, RCVD_IN_DNSWL_MED=-4, SARE_MILLIONSOF=0.315] Received: from mail.ietf.org ([12.22.58.30]) by localhost (ietfa.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id wmpWjvJvB9T5 for ; Wed, 25 Sep 2013 22:55:38 -0700 (PDT) Received: from lhrrgout.huawei.com (lhrrgout.huawei.com [194.213.3.17]) by ietfa.amsl.com (Postfix) with ESMTP id 5E61D11E8140 for ; Wed, 25 Sep 2013 22:55:29 -0700 (PDT) Received: from 172.18.7.190 (EHLO lhreml204-edg.china.huawei.com) ([172.18.7.190]) by lhrrg02-dlp.huawei.com (MOS 4.3.5-GA FastPath queued) with ESMTP id AVW33700; Thu, 26 Sep 2013 05:55:28 +0000 (GMT) Received: from LHREML405-HUB.china.huawei.com (10.201.5.242) by lhreml204-edg.china.huawei.com (172.18.7.223) with Microsoft SMTP Server (TLS) id 14.3.146.0; Thu, 26 Sep 2013 06:54:19 +0100 Received: from SZXEML457-HUB.china.huawei.com (10.82.67.200) by lhreml405-hub.china.huawei.com (10.201.5.242) with Microsoft SMTP Server (TLS) id 14.3.146.0; Thu, 26 Sep 2013 06:55:07 +0100 Received: from szxeml557-mbx.china.huawei.com ([169.254.5.217]) by szxeml457-hub.china.huawei.com ([10.82.67.200]) with mapi id 14.03.0146.000; Thu, 26 Sep 2013 13:55:04 +0800 From: h chan To: "Sri Gundavelli (sgundave)" , "dmm@ietf.org" Thread-Topic: [DMM] I-D Action: draft-ietf-dmm-requirements-07.txt Thread-Index: AQHOogjoa3/L9muNGkS5wozrKtEQEpnXRZkQgABo14CAAAX5UA== Date: Thu, 26 Sep 2013 05:55:03 +0000 Message-ID: <6E31144C030982429702B11D6746B98C370CBC22@szxeml557-mbx.china.huawei.com> References: <1377486056.79906.YahooMailNeo@web163805.mail.gq1.yahoo.com> <24C0F3E22276D9438D6F366EB89FAEA8116A4D4F@xmb-aln-x03.cisco.com> Accept-Language: en-US, zh-CN Content-Language: en-US X-MS-Has-Attach: X-MS-TNEF-Correlator: x-originating-ip: [10.46.81.5] Content-Type: multipart/alternative; boundary="_000_6E31144C030982429702B11D6746B98C370CBC22szxeml557mbxchi_" MIME-Version: 1.0 X-CFilter-Loop: Reflected Subject: Re: [DMM] I-D Action: draft-ietf-dmm-requirements-07.txt X-BeenThere: dmm@ietf.org X-Mailman-Version: 2.1.12 Precedence: list List-Id: Distributed Mobility Management Working Group List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Thu, 26 Sep 2013 05:55:43 -0000 --_000_6E31144C030982429702B11D6746B98C370CBC22szxeml557mbxchi_ Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: quoted-printable Thanks for the comments. We are replying under different sections in separa= te emails. More emails will follow. ----------------------------------------------- We accept your comment to Section 2 and added the explanation of a few term= s there in version 08. Please check. 2.1. Terminology All the general mobility-related terms and their acronyms used in this document are to be interpreted as defined in the Mobile IPv6 base specification [RFC6275], in the Proxy mobile IPv6 specification [RFC5213], and in Mobility Related Terminology [RFC3753]. These terms include the following: mobile node (MN), correspondent node (CN), and home agent (HA) as per [RFC6275]; local mobility anchor (LMA) and mobile access gateway (MAG) as per [RFC5213], and context as per [RFC3753]. In addition, this draft introduces the following terms. Centrally deployed mobility anchors refer to the mobility management deployments in which there are very few mobility anchors and the traffic of millions of mobile nodes in an operator network are managed by the same anchor. Centralized mobility management makes use of centrally deployed mobility anchors. Distributed mobility management is not centralized so that traffic does not need to traverse centrally deployed mobility anchors. Flat mobile network has few levels of routing hierarchy introduced into the data path by the mobility management system. Mobility context is the collection of information required to provide mobility management support for a given mobile node. The following have contributed to the revisions in draft 08: Pierrick, Dape= ng, and Jouni From: dmm-bounces@ietf.org [mailto:dmm-bounces= @ietf.org] On Behalf Of Sri Gundavelli (sgundave) Sent: Sunday, August 25, 2013 10:04 PM To: dmm@ietf.org Subject: Re: [DMM] I-D Action: draft-ietf-dmm-requirements-07.txt Please see inline for some comments. Regards Sri 2. Conventions used in this document 2.1. Terminology All the general mobility-related terms and their acronyms used in this document are to be interpreted as defined in the Mobile IPv6 base specification [RFC6275], in the Proxy mobile IPv6 specification [RFC5213], and in Mobility Related Terminology [RFC3753]. These terms include the following: mobile node (MN), correspondent node (CN), and home agent (HA) as per [RFC6275]; local mobility anchor (LMA) and mobile access gateway (MAG) as per [RFC5213], and context as per [RFC3753]. In addition, this draft introduces the following term. Mobility context is the collection of information required to provide mobility management support for a given mobile node. [Sri] There needs to be some definition of "centrally deployed mobility anc= hor". This term is used through-out the document. What is central, what is = local ? Even in the so called, "centralized anchor models", a gateway can b= e enabled locally. Ex: LMA/MAG, PGW/SGW functions can exist on the same nod= e. Accept and added explanation of this term. Also added the explanation of a = few other terms. --_000_6E31144C030982429702B11D6746B98C370CBC22szxeml557mbxchi_ Content-Type: text/html; charset="us-ascii" Content-Transfer-Encoding: quoted-printable

 <= /p>

Thanks for the comments. = We are replying under different sections in separate emails. More emails wi= ll follow.

-------------------------= ----------------------

We accept your comment to Section 2 and added the explanation of= a few terms there in version 08. Please check.

 <= /p>

2.1.  Terminology

 

   All the general mobility-relate= d terms and their acronyms used in

   this document are to be interpr= eted as defined in the Mobile IPv6

   base specification [RFC6275], i= n the Proxy mobile IPv6 specification

   [RFC5213], and in Mobility Rela= ted Terminology [RFC3753].  These

   terms include the following: mo= bile node (MN), correspondent node

   (CN), and home agent (HA) as pe= r [RFC6275]; local mobility anchor

   (LMA) and mobile access gateway= (MAG) as per [RFC5213], and context

   as per [RFC3753].

 

   In addition, this draft introdu= ces the following terms.

 

   Centrally deployed mobility anc= hors

      refer to the = mobility management deployments in which there are

      very few mobi= lity anchors and the traffic of millions of mobile

      nodes in an o= perator network are managed by the same anchor.

 

   Centralized mobility management=

      makes use of = centrally deployed mobility anchors.

 

   Distributed mobility management=

      is not centra= lized so that traffic does not need to traverse

      centrally dep= loyed mobility anchors.

 

   Flat mobile network<= /span>

      has few level= s of routing hierarchy introduced into the data path

      by the mobili= ty management system.

 

   Mobility context

      is the collec= tion of information required to provide mobility

      management su= pport for a given mobile node.

 <= /p>

The following have contri= buted to the revisions in draft 08: Pierrick, Dapeng, and Jouni<= /span>

 <= /p>

 <= /p>

From: dmm-bounces@ietf.org [mailto:dmm-bounces@ietf.org] On Behalf Of Sri Gundavelli (sgundave)
Sent: Sunday, August 25, 2013 10:04 PM
To: dmm@ietf.org
Subject: Re: [DMM] I-D Action: draft-ietf-dmm-requirements-07.txt

 

Please see inline for some comments.

 

Regards

Sri

 

2.  Convention=
s used in this document
 
2.1.  Terminol=
ogy
 
   All th=
e general mobility-related terms and their acronyms used in
   this d=
ocument are to be interpreted as defined in the Mobile IPv6
   base s=
pecification [RFC6275], in the Proxy mobile IPv6 specification
   [RFC52=
13], and in Mobility Related Terminology [RFC3753].  These<=
/span>
   terms =
include the following: mobile node (MN), correspondent node
   (CN), =
and home agent (HA) as per [RFC6275]; local mobility anchor
   (LMA) =
and mobile access gateway (MAG) as per [RFC5213], and context
   as per=
 [RFC3753].
 
   In add=
ition, this draft introduces the following term.
 
   Mobili=
ty context
 
   &=
nbsp;  is the collection of information required to provide mobility
   &=
nbsp;  management support for a given mobile node.

 


 


[Sri] There needs=
 to be some definition of "centrally deployed mobility anchor". T=
his term is used through-out the document. What is central, what is local ?=
 Even in the so called, "centralized anchor models", a gateway ca=
n be enabled locally. Ex: LMA/MAG, PGW/SGW functions can exist on the same =
node. 


 

Accept and added explanation of this term. Also added= the explanation of a few other terms.

--_000_6E31144C030982429702B11D6746B98C370CBC22szxeml557mbxchi_-- From h.anthony.chan@huawei.com Wed Sep 25 23:08:46 2013 Return-Path: X-Original-To: dmm@ietfa.amsl.com Delivered-To: dmm@ietfa.amsl.com Received: from localhost (localhost [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id F1D4511E8146 for ; Wed, 25 Sep 2013 23:08:45 -0700 (PDT) X-Virus-Scanned: amavisd-new at amsl.com X-Spam-Flag: NO X-Spam-Score: -6.441 X-Spam-Level: X-Spam-Status: No, score=-6.441 tagged_above=-999 required=5 tests=[AWL=0.157, BAYES_00=-2.599, HTML_MESSAGE=0.001, RCVD_IN_DNSWL_MED=-4] Received: from mail.ietf.org ([12.22.58.30]) by localhost (ietfa.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id vuMNp05IOvY0 for ; Wed, 25 Sep 2013 23:08:39 -0700 (PDT) Received: from lhrrgout.huawei.com (lhrrgout.huawei.com [194.213.3.17]) by ietfa.amsl.com (Postfix) with ESMTP id 82D8521F9A64 for ; Wed, 25 Sep 2013 23:08:35 -0700 (PDT) Received: from 172.18.7.190 (EHLO lhreml203-edg.china.huawei.com) ([172.18.7.190]) by lhrrg02-dlp.huawei.com (MOS 4.3.5-GA FastPath queued) with ESMTP id AVW34730; Thu, 26 Sep 2013 06:08:34 +0000 (GMT) Received: from LHREML401-HUB.china.huawei.com (10.201.5.240) by lhreml203-edg.huawei.com (172.18.7.221) with Microsoft SMTP Server (TLS) id 14.3.146.0; Thu, 26 Sep 2013 07:07:38 +0100 Received: from SZXEML405-HUB.china.huawei.com (10.82.67.60) by lhreml401-hub.china.huawei.com (10.201.5.240) with Microsoft SMTP Server (TLS) id 14.3.146.0; Thu, 26 Sep 2013 07:08:30 +0100 Received: from szxeml557-mbx.china.huawei.com ([169.254.5.217]) by szxeml405-hub.china.huawei.com ([::1]) with mapi id 14.03.0146.000; Thu, 26 Sep 2013 14:08:25 +0800 From: h chan To: "Sri Gundavelli (sgundave)" , "dmm@ietf.org" Thread-Topic: [DMM] I-D Action: draft-ietf-dmm-requirements-07.txt Thread-Index: AQHOogjoa3/L9muNGkS5wozrKtEQEpnXRZkQgABo14CAAAeAIA== Date: Thu, 26 Sep 2013 06:08:24 +0000 Message-ID: <6E31144C030982429702B11D6746B98C370CBC3F@szxeml557-mbx.china.huawei.com> References: <1377486056.79906.YahooMailNeo@web163805.mail.gq1.yahoo.com> <24C0F3E22276D9438D6F366EB89FAEA8116A4D4F@xmb-aln-x03.cisco.com> Accept-Language: en-US, zh-CN Content-Language: en-US X-MS-Has-Attach: X-MS-TNEF-Correlator: x-originating-ip: [10.46.81.5] Content-Type: multipart/alternative; boundary="_000_6E31144C030982429702B11D6746B98C370CBC3Fszxeml557mbxchi_" MIME-Version: 1.0 X-CFilter-Loop: Reflected Subject: Re: [DMM] I-D Action: draft-ietf-dmm-requirements-07.txt X-BeenThere: dmm@ietf.org X-Mailman-Version: 2.1.12 Precedence: list List-Id: Distributed Mobility Management Working Group List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Thu, 26 Sep 2013 06:08:46 -0000 --_000_6E31144C030982429702B11D6746B98C370CBC3Fszxeml557mbxchi_ Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: quoted-printable Sri, Thanks for the comments. We are replying under different sections in separa= te emails. More emails will follow. ----------------------------------------------- We have revised Section 3 in version 08. Our responses are in-line. Please = check. 3. Centralized versus distributed mobility management Mobility management functions may be implemented at different layers of the protocol stack. At the IP (network) layer, mobility management can be client-based or network-based. An IP-layer mobility management protocol is typically based on the principle of distinguishing between session identifier and routing address and maintaining a mapping between the two. In Mobile IP, the home address serves as the session identifier whereas the care-of- address (CoA) takes the role of the routing address. The binding between these two is maintained at the home agent (mobility anchor). If packets addressed to the home address of a mobile node can be continuously delivered to the node, then all sessions using that home address are unaffected even though the routing address (CoA) changes. The next two subsections explain centralized and distributed mobility management functions in the network. 3.1. Centralized mobility management In centralized mobility management, the mapping information between the session identifier and the locator IP address of a mobile node (MN) is kept at a single mobility anchor. At the same time, packets destined to the MN are routed via this anchor. In other words, such mobility management systems are centralized in both the control plane and the data plane (mobile node IP traffic). Many existing mobility management deployments make use of centralized mobility anchoring in a hierarchical network architecture, as shown in Figure 1. Examples of such centralized mobility anchors are the home agent (HA) and local mobility anchor (LMA) in Mobile IPv6 [RFC6275] and Proxy Mobile IPv6 [RFC5213], respectively. Current cellular networks such as the Third Generation Partnership Project (3GPP) GPRS networks, CDMA networks, and 3GPP Evolved Packet System (EPS) networks employ centralized mobility management too. In particular, the Gateway GPRS Support Node (GGSN), Serving GPRS Support Node (SGSN) and Radio Network Controller (RNC) in the 3GPP GPRS hierarchical network, and the Packet Data Network Gateway (P-GW) and Serving Gateway (S-GW) in the 3GPP EPS network all act as anchors in a hierarchy. 3G GPRS 3GPP EPS MIP/PMIP +------+ +------+ +------+ | GGSN | | P-GW | |HA/LMA| +------+ +------+ +------+ /\ /\ /\ / \ / \ / \ / \ / \ / \ / \ / \ / \ / \ / \ / \ / \ / \ / \ / \ / \ / \ +------+ +------+ +------+ +------+ +------+ +------+ | SGSN | | SGSN | | S-GW | | S-GW | |MN/MAG| |MN/MAG| +------+ +------+ +------+ +------+ +------+ +------+ /\ /\ / \ / \ / \ / \ +---+ +---+ +---+ +---+ |RNC| |RNC| |RNC| |RNC| +---+ +---+ +---+ +---+ Figure 1. Centralized mobility management. 3.2. Distributed mobility management Mobility management functions may also be distributed to multiple networks as shown in Figure 2, so that a mobile node in any of these networks may be served by a nearby mobility function (MF). +------+ +------+ +------+ +------+ | MF | | MF | | MF | | MF | +------+ +------+ +------+ +------+ | +----+ | MN | +----+ Figure 2. Distributed mobility management. Mobility management may be partially or fully distributed [I-D.yokota-dmm-scenario]. In the former case only the data plane is distributed, implicitly assuming separation of data and control planes as described in [I-D.wakikawa-netext-pmip-cp-up-separtion]. Fully distributed mobility management implies that both the data plane and the control plane are distributed. While mobility management can be distributed, it is not necessary for other functions such as subscription management, subscription database, and network access authentication to be similarly distributed. A distributed mobility management scheme for a flat mobile network of access nodes is proposed in [Paper-Distributed.Dynamic.Mobility]. Its benefits over centralized mobility management are shown through simulations in [Paper-Distributed.Centralized.Mobility]. Moreover, the (re)use and extension of existing protocols in the design of both fully distributed mobility management [Paper-Migrating.Home.Agents] [Paper-Distributed.Mobility.SAE] and partially distributed mobility management [Paper-Distributed.Mobility.PMIP] [Paper- Distributed.Mobility.MIP] have been reported in the literature. Therefore, before designing new mobility management protocols for a future distributed architecture, it is recommended to first consider whether existing mobility management protocols can be extended. The following have contributed to the revisions in draft 08: Pierrick, Dape= ng, and Jouni From: dmm-bounces@ietf.org [mailto:dmm-bounces= @ietf.org] On Behalf Of Sri Gundavelli (sgundave) Sent: Sunday, August 25, 2013 10:04 PM To: dmm@ietf.org Subject: Re: [DMM] I-D Action: draft-ietf-dmm-requirements-07.txt Please see inline for some comments. Regards Sri 3. Centralized versus distributed mobility management Mobility management functions may be implemented at different layers of the protocol stack. At the IP (network) layer, they may reside in the network or in the mobile node. In particular, a network-based solution resides in the network only. It therefore enables mobility for existing hosts and network applications which are already in deployment but lack mobility support. [Sri] Above text is bit confusing to me, specially the last sentence. Mobil= ity management can be based on client-based, or network-based. Accept and revised in version 08 At the IP layer, a mobility management protocol supporting session continuity is typically based on the principle of distinguishing between identifier and routing address and maintaining a mapping between the two. [Sri] Replace, "Session continuity" with "IP mobility" or "IP address conti= nuity". Delete session continuity, but also keep the discussion general and not spe= cific to MIP. In Mobile IP, the home address serves as an identifier of the device whereas the care-of-address (CoA) takes the role of the routing address. The binding between these two is maintained at the home agent (mobility anchor). If packets can be continuously delivered to a mobile node at its home address, then all sessions using that home address are unaffected even though the routing address (CoA) changes. [Sri] We should leave it at, "IP address mobility". Revised this section but tried to keep it general and not specific to MIP. The next two subsections explain centralized and distributed mobility management functions in the network. 3.1. Centralized mobility management In centralized mobility management, the mapping information between the persistent node identifier and the locator IP address of a mobile node (MN) is kept at a single mobility anchor. At the same time, packets destined to the MN are routed via this anchor. [Sri] Can we use the MIP terminology, home address/Care-of address terminol= ogy, as supposed to LISP terminology ? Revised this section but tried to keep this section general and not specifi= c to MIP In other words, such mobility management systems are centralized in both the control plane and the data plane (mobile node IP traffic). Many existing mobility management deployments make use of centralized mobility anchoring in a hierarchical network architecture, as shown in Figure 1. Examples of such centralized mobility anchors are the home agent (HA) and local mobility anchor (LMA) in Mobile IPv6 [RFC6275] and Proxy Mobile IPv6 [RFC5213], respectively. Current cellular networks such as the Third Generation Partnership Project (3GPP) GPRS networks, CDMA networks, and 3GPP Evolved Packet System (EPS) networks employ centralized mobility management too. In particular, the Gateway GPRS Support Node (GGSN), Serving GPRS Support Node (SGSN) and Radio Network Controller (RNC) in the 3GPP GPRS hierarchical network, and the Packet Data Network Gateway (P-GW) and Serving Gateway (S-GW) in the 3GPP EPS network all act as anchors in a hierarchy. Chan (Ed.), et al. Expires February 3, 2014 [Page 7] Internet-Draft DMM-Reqs August 2013 3G GPRS 3GPP EPS MIP/PMIP +------+ +------+ +------+ | GGSN | | P-GW | |HA/LMA| +------+ +------+ +------+ /\ /\ /\ / \ / \ / \ / \ / \ / \ / \ / \ / \ / \ / \ / \ / \ / \ / \ / \ / \ / \ +------+ +------+ +------+ +------+ +------+ +------+ | SGSN | | SGSN | | S-GW | | S-GW | |MN/MAG| |MN/MAG| +------+ +------+ +------+ +------+ +------+ +------+ /\ /\ / \ / \ / \ / \ +---+ +---+ +---+ +---+ |RNC| |RNC| |RNC| |RNC| +---+ +---+ +---+ +---+ Figure 1. Centralized mobility management. 3.2. Distributed mobility management Mobility management functions may also be distributed to multiple networks as shown in Figure 2, so that a mobile node in any of these networks may be served by a nearby mobility function (MF). +------+ +------+ +------+ +------+ | MF | | MF | | MF | | MF | +------+ +------+ +------+ +------+ | +----+ | MN | +----+ Figure 2. Distributed mobility management. Mobility management may be partially or fully distributed. In the former case only the data plane is distributed. Fully distributed mobility management implies that both the data plane and the control plane are distributed. Such concepts of data and control plane separation are not yet described in the IETF developed mobility protocols so far but are described in detail in [I-D.yokota-dmm- scenario]. While mobility management can be distributed, it is not necessary for other functions such as subscription management, Chan (Ed.), et al. Expires February 3, 2014 [Page 8] Internet-Draft DMM-Reqs August 2013 subscription database, and network access authentication to be similarly distributed. [Sri] The case of centralized CP and distributed DP is covered in IETF docs= , http://datatracker.ietf.org/doc/draft-wakikawa-netext-pmip-cp-up-separation= / This is a variant of the DMM models. Already quoted the yokota reference but added this reference anyway. A distributed mobility management scheme for flat IP-based mobile network architecture consisting of access nodes is proposed in [Paper-Distributed.Dynamic.Mobility]. Its benefits over centralized mobility management are shown through simulations in [Paper- Distributed.Centralized.Mobility]. Moreover, the (re)use and extension of existing protocols in the design of both fully distributed mobility management [Paper-Migrating.Home.Agents] [Paper- Distributed.Mobility.SAE] and partially distributed mobility management [Paper-Distributed.Mobility.PMIP] [Paper- Distributed.Mobility.MIP] have been reported in the literature. Therefore, before designing new mobility management protocols for a future flat IP architecture, it is recommended to first consider whether existing mobility management protocols can be extended to serve a flat IP architecture. [Sri] Lot of unnecessary text in this document. Not sure, we need all of th= is text. Added the term "flat mobile network" and revised in version 08 --_000_6E31144C030982429702B11D6746B98C370CBC3Fszxeml557mbxchi_ Content-Type: text/html; charset="us-ascii" Content-Transfer-Encoding: quoted-printable

Sri,

 <= /p>

Thanks for the comments. = We are replying under different sections in separate emails. More emails wi= ll follow.

-------------------------= ----------------------

We have revised Section 3 in version 08. Our responses are in-line. Please check.

 <= /p>

3.  Centralized versus distributed mobi= lity management

 

   Mobility management functions m= ay be implemented at different layers

   of the protocol stack.  At= the IP (network) layer, mobility

   management can be client-based = or network-based.

 

   An IP-layer mobility management= protocol is typically based on the

   principle of distinguishing bet= ween session identifier and routing

   address and maintaining a mappi= ng between the two.  In Mobile IP, the

   home address serves as the sess= ion identifier whereas the care-of-

   address (CoA) takes the role of= the routing address.  The binding

   between these two is maintained= at the home agent (mobility anchor).

   If packets addressed to the hom= e address of a mobile node can be

   continuously delivered to the n= ode, then all sessions using that home

   address are unaffected even tho= ugh the routing address (CoA) changes.

 

   The next two subsections explai= n centralized and distributed mobility

   management functions in the net= work.

 

3.1.  Centralized mobility management

 

   In centralized mobility managem= ent, the mapping information between

   the session identifier and the = locator IP address of a mobile node

   (MN) is kept at a single mobili= ty anchor.  At the same time, packets

   destined to the MN are routed v= ia this anchor.  In other words, such

   mobility management systems are= centralized in both the control plane

   and the data plane (mobile node= IP traffic).

 

   Many existing mobility manageme= nt deployments make use of centralized

   mobility anchoring in a hierarc= hical network architecture, as shown

   in Figure 1.  Examples of = such centralized mobility anchors are the

   home agent (HA) and local mobil= ity anchor (LMA) in Mobile IPv6

   [RFC6275] and Proxy Mobile IPv6= [RFC5213], respectively.  Current

   cellular networks such as the T= hird Generation Partnership Project

   (3GPP) GPRS networks, CDMA netw= orks, and 3GPP Evolved Packet System

   (EPS) networks employ centraliz= ed mobility management too.  In

   particular, the Gateway GPRS Su= pport Node (GGSN), Serving GPRS

   Support Node (SGSN) and Radio N= etwork Controller (RNC) in the 3GPP

   GPRS hierarchical network, and = the Packet Data Network Gateway (P-GW)

   and Serving Gateway (S-GW) in t= he 3GPP EPS network all act as anchors

   in a hierarchy.

 

 

       &n= bsp; 3G GPRS          &nb= sp;      3GPP EPS     &nb= sp;          MIP/PMIP

       &n= bsp; +------+         =        +------+   &nbs= p;            +-= -----+

       &n= bsp; | GGSN |          &n= bsp;     | P-GW |      &n= bsp;         |HA/LMA|

       &n= bsp; +------+         =        +------+   &nbs= p;            +-= -----+

       &n= bsp;    /\        &n= bsp;            = ; /\            &nbs= p;         /\

       &n= bsp;   /  \        &= nbsp;           /  \=             &nb= sp;       /  \

       &n= bsp;  /    \       &= nbsp;          /  &n= bsp; \           &nb= sp;      /    \

       &n= bsp; /      \      &= nbsp;         /   &n= bsp;  \          &nb= sp;     /      \

        /=         \     &= nbsp;        /    &n= bsp;   \          &n= bsp;   /        \

       / =          \    &= nbsp;       /     &n= bsp;    \        &nb= sp;   /          \

      /  =           \   &= nbsp;      /      &n= bsp;     \       &nb= sp;  /          &nbs= p; \

  +------+   &nb= sp;  +------+  +------+    &n= bsp; +------+  +------+      = +------+

  | SGSN |     = ; | SGSN |  | S-GW |      | S-GW |  |MN/= MAG|      |MN/MAG|

  +------+   &nb= sp;  +------+  +------+    &n= bsp; +------+  +------+      = +------+

     /\   = ;         /\

    /  \  &nbs= p;       /  \

   /    \ &nbs= p;      /    \

+---+  +---+  += ;---+  +---+

|RNC|  |RNC|  |RNC|  |RNC|

+---+  +---+  += ;---+  +---+

 

   Figure 1.  Centralized mob= ility management.

 

3.2.  Distributed mobility management

 

   Mobility management functions m= ay also be distributed to multiple

   networks as shown in Figure 2, = so that a mobile node in any of these

   networks may be served by a nea= rby mobility function (MF).

 

 

       &n= bsp;            += ;------+  +------+  +------+  +-----= -+

       &n= bsp;            |&nb= sp; MF  |  |  MF  |  |  MF  |  |&nb= sp; MF  |

       &n= bsp;            += ;------+  +------+  +------+  +-----= -+

       &n= bsp;            &nbs= p;            &= nbsp;         |

       &n= bsp;            = ;            &n= bsp;        +----+

       &n= bsp;            = ;            &n= bsp;        | MN |

       &n= bsp;            = ;            &n= bsp;        +----+

 

   Figure 2.  Distributed mob= ility management.

 

   Mobility management may be part= ially or fully distributed

   [I-D.yokota-dmm-scenario]. = ; In the former case only the data plane is

   distributed, implicitly assumin= g separation of data and control

   planes as described in [I-D.wak= ikawa-netext-pmip-cp-up-separtion].

   Fully distributed mobility mana= gement implies that both the data

   plane and the control plane are= distributed.  While mobility

   management can be distributed, = it is not necessary for other

   functions such as subscription = management, subscription database, and

   network access authentication t= o be similarly distributed.

 

   A distributed mobility manageme= nt scheme for a flat mobile network of

   access nodes is proposed in [Pa= per-Distributed.Dynamic.Mobility].

   Its benefits over centralized m= obility management are shown through

   simulations in [Paper-Distribut= ed.Centralized.Mobility].  Moreover,

   the (re)use and extension of ex= isting protocols in the design of both

   fully distributed mobility mana= gement [Paper-Migrating.Home.Agents]

   [Paper-Distributed.Mobility.SAE= ] and partially distributed mobility

   management [Paper-Distributed.M= obility.PMIP] [Paper-

   Distributed.Mobility.MIP] have = been reported in the literature.

   Therefore, before designing new= mobility management protocols for a

   future distributed architecture= , it is recommended to first consider

   whether existing mobility manag= ement protocols can be extended.

 

 <= /p>

The following have contri= buted to the revisions in draft 08: Pierrick, Dapeng, and Jouni<= /span>

 <= /p>

 <= /p>

From: dmm-bounces@ietf.org [mailto:dmm-bounces@ietf.org] On Behalf Of Sri Gundavelli (sgundave)
Sent: Sunday, August 25, 2013 10:04 PM
To: dmm@ietf.org
Subject: Re: [DMM] I-D Action: draft-ietf-dmm-requirements-07.txt

 

Please see inline for some comments.

 

Regards

Sri

 =

 
3.  Centralize=
d versus distributed mobility management
 
   Mobili=
ty management functions may be implemented at different layers
   of the=
 protocol stack.  At the IP (network) layer, they may reside in
   the ne=
twork or in the mobile node.  In particular, a network-based
   soluti=
on resides in the network only.  It therefore enables mobility
   for ex=
isting hosts and network applications which are already in
   deploy=
ment but lack mobility support.
 
[Sri] Above text =
is bit confusing to me, specially the last sentence. Mobility management ca=
n be based on client-based, or network-based. 
 =

Accept and revised i=
n version 08
 
   At the=
 IP layer, a mobility management protocol supporting session
   cont=
inuity is typically based on the principle o=
f distinguishing
   betwee=
n identifier and routing address and maintaining a mapping
   betwee=
n the two.  
[Sri] Replace, &q=
uot;Session continuity" with "IP mobility" or "IP addre=
ss continuity". 
 =

Delete session conti=
nuity, but also keep the discussion general and not specific to MIP.
 
In Mobile IP, the h=
ome address serves as an
   identi=
fier of the device whereas the care-of-address (CoA) takes the
   role o=
f the routing address.  The binding between these two is
   mainta=
ined at the home agent (mobility anchor).  If packets can be
   contin=
uously delivered to a mobile node at its home address, then all<=
/span>
   sessio=
ns using that home address are unaffected even though the=

   routin=
g address (CoA) changes.
 
[Sri] We should l=
eave it at, "IP address mobility".
 =

Revised this section=
 but tried to keep it general and not specific to MIP.

 


   The ne=
xt two subsections explain centralized and distributed mobility<=
/span>


   manage=
ment functions in the network.


 


3.1.  Centrali=
zed mobility management


 


   In cen=
tralized mobility management, the mapping information between


   the pe=
rsistent node identifier and the locator IP address of a mobile<=
/span>


   node (=
MN) is kept at a single mobility anchor.  At the same time,=



   packet=
s destined to the MN are routed via this anchor.


[Sri] Can we use =
the MIP terminology, home address/Care-of address terminology, as supposed =
to LISP terminology ? 


 


Revised this section=
 but tried to keep this section general and not specific to MIP<=
/span>


  In other


   words,=
 such mobility management systems are centralized in both the


   contro=
l plane and the data plane (mobile node IP traffic).

 


   Many e=
xisting mobility management deployments make use of centralized<=
/span>


   mobili=
ty anchoring in a hierarchical network architecture, as shown


   in Fig=
ure 1.  Examples of such centralized mobility anchors are the


   home a=
gent (HA) and local mobility anchor (LMA) in Mobile IPv6<=
/pre>

   [RFC62=
75] and Proxy Mobile IPv6 [RFC5213], respectively.  Current=



   cellul=
ar networks such as the Third Generation Partnership Project


   (3GPP)=
 GPRS networks, CDMA networks, and 3GPP Evolved Packet System


   (EPS) =
networks employ centralized mobility management too.  In


   partic=
ular, the Gateway GPRS Support Node (GGSN), Serving GPRS<=
/pre>

   Suppor=
t Node (SGSN) and Radio Network Controller (RNC) in the 3GPP


   GPRS h=
ierarchical network, and the Packet Data Network Gateway (P-GW)<=
/span>


   and Se=
rving Gateway (S-GW) in the 3GPP EPS network all act as anchors<=
/span>


   in a h=
ierarchy.


 


 


 


 


 


 


 


 


 


 


 


 


Chan (Ed.), et al.&=
nbsp;     Expires February 3, 2014   &nb=
sp;            [Page=
 7]


 


Internet-Draft =
;            &n=
bsp;    DMM-Reqs       &n=
bsp;            =
; August 2013


 


 


   &=
nbsp;     3G GPRS      &n=
bsp;          3GPP EPS &n=
bsp;            =
;  MIP/PMIP


   &=
nbsp;     +------+     =
;           +------&#=
43;            =
    +------+


   &=
nbsp;     | GGSN |      &=
nbsp;         | P-GW |  &=
nbsp;           &nbs=
p; |HA/LMA|


   &=
nbsp;     +------+     =
;           +------&#=
43;            =
    +------+


   &=
nbsp;        /\    &=
nbsp;            &nb=
sp;    /\       &nbs=
p;            &=
nbsp; /\


   &=
nbsp;       /  \    =
            &nb=
sp;   /  \        &n=
bsp;           /  \<=
o:p>


   &=
nbsp;      /    \   =
            &nb=
sp;  /    \       &n=
bsp;          /  &nb=
sp; \


   &=
nbsp;     /      \  =
            &nb=
sp; /      \      &n=
bsp;         /   &nb=
sp;  \


   &=
nbsp;    /        \ =
             /&=
nbsp;       \     &n=
bsp;        /    &nb=
sp;   \


   &=
nbsp;   /          \=
            / &=
nbsp;        \    &n=
bsp;       /     &nb=
sp;    \


   &=
nbsp;  /          &n=
bsp; \          /  &=
nbsp;         \   &n=
bsp;      /      &nb=
sp;     \


  +------&=
#43;      +------+  +------+&=
nbsp;     +------+  +------+ =
     +------+


  | SGSN |&nbs=
p;     | SGSN |  | S-GW |    &=
nbsp; | S-GW |  |MN/MAG|      |MN/MAG|


  +------&=
#43;      +------+  +------+&=
nbsp;     +------+  +------+ =
     +------+


   &=
nbsp; /\            =
/\


    =
/  \          /  \


   / =
;   \        /  &nbs=
p; \


+---+ =
 +---+  +---+  +---+

|RNC|  |RNC|&n=
bsp; |RNC|  |RNC|


+---+ =
 +---+  +---+  +---+

 


   Figure=
 1.  Centralized mobility management.


 


3.2.  Distribu=
ted mobility management


 


   Mobili=
ty management functions may also be distributed to multiple


   networ=
ks as shown in Figure 2, so that a mobile node in any of these


   networ=
ks may be served by a nearby mobility function (MF).

 


 


   &=
nbsp;           &nbs=
p;    +------+  +------+  +---=
---+  +------+


   &=
nbsp;           &nbs=
p;    |  MF  |  |  MF  |  |&nb=
sp; MF  |  |  MF  |


   &=
nbsp;           &nbs=
p;    +------+  +------+  +---=
---+  +------+


    =
            &nb=
sp;            =
            &nb=
sp; |


   &=
nbsp;           &nbs=
p;            &=
nbsp;            =
3;----+


   &=
nbsp;           &nbs=
p;            &=
nbsp;            | M=
N |


   &=
nbsp;           &nbs=
p;            &=
nbsp;            =
3;----+


 


   Figure=
 2.  Distributed mobility management.


 


 


   Mobili=
ty management may be partially or fully distributed.  In the


   former=
 case only the data plane is distributed.  Fully distributed


   mobili=
ty management implies that both the data plane and the control


   plane =
are distributed.  Such concepts of data and control plane


   separa=
tion are not yet described in the IETF developed mobility=



   protoc=
ols so far but are described in detail in [I-D.yokota-dmm-


   scenar=
io].  While mobility management can be distributed, it is not


   necess=
ary for other functions such as subscription management,<=
/pre>

 


 


 


Chan (Ed.), et al.&=
nbsp;     Expires February 3, 2014   &nb=
sp;            [Page=
 8]


 


Internet-Draft =
;            &n=
bsp;    DMM-Reqs       &n=
bsp;            =
; August 2013


 


 


   subscr=
iption database, and network access authentication to be<=
/pre>

   simila=
rly distributed.


 


[Sri] The case of centralized CP =
and distributed DP is covered in IETF docs, 


ht=
tp://datatracker.ietf.org/doc/draft-wakikawa-netext-pmip-cp-up-separation/<=
/a>


This is a varian=
t of the DMM models. <=
/pre>

 


Already quoted the yo=
kota reference but added this reference anyway.


 


   A dist=
ributed mobility management scheme for flat IP-based mobile


   networ=
k architecture consisting of access nodes is proposed in<=
/pre>

   [Paper=
-Distributed.Dynamic.Mobility].  Its benefits over centralized


   mobili=
ty management are shown through simulations in [Paper-

   Distri=
buted.Centralized.Mobility].  Moreover, the (re)use and


   extens=
ion of existing protocols in the design of both fully

   distri=
buted mobility management [Paper-Migrating.Home.Agents] [Paper-<=
/span>


   Distri=
buted.Mobility.SAE] and partially distributed mobility

   manage=
ment [Paper-Distributed.Mobility.PMIP] [Paper-


   Distri=
buted.Mobility.MIP] have been reported in the literature.=



   Theref=
ore, before designing new mobility management protocols for a


   future=
 flat IP architecture, it is recommended to first consider


   whethe=
r existing mobility management protocols can be extended to


   serve =
a flat IP architecture.


 


[Sri] =
Lot of unnecessary text in this document. Not sure, we need all o=
f this text.


 


Added the term “=
;flat mobile network” and revised in version 08
--_000_6E31144C030982429702B11D6746B98C370CBC3Fszxeml557mbxchi_-- From h.anthony.chan@huawei.com Thu Sep 26 08:22:00 2013 Return-Path: X-Original-To: dmm@ietfa.amsl.com Delivered-To: dmm@ietfa.amsl.com Received: from localhost (localhost [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id D904811E80D2 for ; Thu, 26 Sep 2013 08:21:59 -0700 (PDT) X-Virus-Scanned: amavisd-new at amsl.com X-Spam-Flag: NO X-Spam-Score: -6.493 X-Spam-Level: X-Spam-Status: No, score=-6.493 tagged_above=-999 required=5 tests=[AWL=0.105, BAYES_00=-2.599, HTML_MESSAGE=0.001, RCVD_IN_DNSWL_MED=-4] Received: from mail.ietf.org ([12.22.58.30]) by localhost (ietfa.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id 76YI99pozUeq for ; Thu, 26 Sep 2013 08:21:44 -0700 (PDT) Received: from lhrrgout.huawei.com (lhrrgout.huawei.com [194.213.3.17]) by ietfa.amsl.com (Postfix) with ESMTP id 7C4C211E8104 for ; Thu, 26 Sep 2013 08:21:36 -0700 (PDT) Received: from 172.18.7.190 (EHLO lhreml203-edg.china.huawei.com) ([172.18.7.190]) by lhrrg02-dlp.huawei.com (MOS 4.3.5-GA FastPath queued) with ESMTP id AVW87561; Thu, 26 Sep 2013 15:21:33 +0000 (GMT) Received: from LHREML404-HUB.china.huawei.com (10.201.5.218) by lhreml203-edg.huawei.com (172.18.7.221) with Microsoft SMTP Server (TLS) id 14.3.146.0; Thu, 26 Sep 2013 16:20:36 +0100 Received: from SZXEML460-HUB.china.huawei.com (10.82.67.203) by lhreml404-hub.china.huawei.com (10.201.5.218) with Microsoft SMTP Server (TLS) id 14.3.146.0; Thu, 26 Sep 2013 16:21:30 +0100 Received: from szxeml557-mbx.china.huawei.com ([169.254.5.217]) by szxeml460-hub.china.huawei.com ([10.82.67.203]) with mapi id 14.03.0146.000; Thu, 26 Sep 2013 23:21:24 +0800 From: h chan To: "Sri Gundavelli (sgundave)" , "dmm@ietf.org" Thread-Topic: [DMM] I-D Action: draft-ietf-dmm-requirements-07.txt Thread-Index: AQHOogjoa3/L9muNGkS5wozrKtEQEpnXRZkQgABo14CAAJoKEA== Date: Thu, 26 Sep 2013 15:21:24 +0000 Message-ID: <6E31144C030982429702B11D6746B98C370CBCD6@szxeml557-mbx.china.huawei.com> References: <1377486056.79906.YahooMailNeo@web163805.mail.gq1.yahoo.com> <24C0F3E22276D9438D6F366EB89FAEA8116A4D4F@xmb-aln-x03.cisco.com> Accept-Language: en-US, zh-CN Content-Language: en-US X-MS-Has-Attach: X-MS-TNEF-Correlator: x-originating-ip: [10.192.11.128] Content-Type: multipart/alternative; boundary="_000_6E31144C030982429702B11D6746B98C370CBCD6szxeml557mbxchi_" MIME-Version: 1.0 X-CFilter-Loop: Reflected Subject: Re: [DMM] I-D Action: draft-ietf-dmm-requirements-07.txt X-BeenThere: dmm@ietf.org X-Mailman-Version: 2.1.12 Precedence: list List-Id: Distributed Mobility Management Working Group List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Thu, 26 Sep 2013 15:22:00 -0000 --_000_6E31144C030982429702B11D6746B98C370CBCD6szxeml557mbxchi_ Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: quoted-printable Sri, We have revised the problem statements in Section 4 in version 08. Our resp= onses are in-line. Please check again. 4. Problem Statement The problems that can be addressed with DMM are summarized in the following: PS1: Non-optimal routes Routing via a centralized anchor often results in non-optimal routes, thereby increasing the end-to-end delay. The problem is manifested, for example, when accessing a local server or servers of a Content Delivery Network (CDN), or when receiving locally available IP multicast or sending IP multicast packets. PS2: Divergence from other evolutionary trends in network architectures such as distribution of content delivery. Centralized mobility management can become non-optimal with a flat network architecture. PS3: Low scalability of centralized tunnel management and mobility context maintenance Setting up tunnels through a central anchor and maintaining mobility context for each MN usually requires more concentrated resources in a centralized design, thus reducing scalability. Distributing the tunnel maintenance function and the mobility context maintenance function among different network entities with proper signaling protocol design can increase scalability. PS4: Single point of failure and attack Centralized anchoring designs may be more vulnerable to single points of failures and attacks than a distributed system. The impact of a successful attack on a system with centralized mobility management can be far greater as well. PS5: Unnecessary mobility support to nodes that do not need it IP mobility support is not always required, and not every parameter of mobility context is always used. For example, some applications do not need a stable IP address during a handover to maintain session continuity. Sometimes, the entire application session runs while the terminal does not change the point of attachment. Besides, some sessions, e.g. SIP-based sessions, can handle mobility at the application layer and hence do not need IP mobility support; it is then more efficient to deactivate IP mobility support for such sessions. PS6: (Related problem) Mobility signaling overhead with peer-to-peer communication Wasting resources when mobility signaling (e.g., maintenance of the tunnel, keep alive signaling, etc.) is not turned off for peer-to-peer communication. Peer-to-peer communications have particular traffic patterns that often do not benefit from mobility support from the network. Thus, the associated mobility support signaling (e.g., maintenance of the tunnel, keep alive signaling, etc.) wastes network resources for no application gain. PS7: (Related problem) Deployment with multiple mobility solutions There are already many variants and extensions of MIP. Deployment of new mobility management solutions can be challenging, and debugging difficult, when they must co-exist with solutions already in the field. PS8: Duplicate multicast traffic IP multicast distribution over architectures using IP mobility solutions (e.g. RFC6224) may lead to convergence of duplicated multicast subscriptions towards the downstream tunnel entity (e.g. MAG in PMIPv6). Concretely, when multicast subscription for individual mobile nodes is coupled with mobility tunnels (e.g. PMIPv6 tunnel), duplicate multicast subscription(s) is prone to be received through different upstream paths. This problem may also exist or be more severe in a distributed mobility environment. The following have contributed to the above revisions in draft 08: Pierrick= , Dapeng, and Jouni From: dmm-bounces@ietf.org [mailto:dmm-bounces= @ietf.org] On Behalf Of Sri Gundavelli (sgundave) Sent: Sunday, August 25, 2013 10:04 PM To: dmm@ietf.org Subject: Re: [DMM] I-D Action: draft-ietf-dmm-requirements-07.txt Please see inline for some comments. Regards Sri 4. Problem Statement The problems that can be addressed with DMM are summarized in the following: PS1: Non-optimal routes Routing via a centralized anchor often results in a longer route. [Sri] "longer route" ? I assume this is about routing/tx delay. Please re-= word. Accept and revised in version 08 The problem is manifested, for example, when accessing a local server or servers of a Content Delivery Network (CDN), or when receiving locally available IP multicast or sending IP multicast packets. [Sri] Does RFC 6705, or RFC 6909 does not address this issue ? May be the C= DN example is incorrect. CDN example is also an important driver: distributing the mobility anchor i= s more compatible with distributing and replicating the contents, which may= be more general to the particular protocol solutions. It may also include = the case that there is a CDN server is in a neighboring network. Suggestion= s to reword (for version 09) are welcome. PS2: Divergence from other evolutionary trends in network architectures such as distribution of content delivery. Centralized mobility management can become non-optimal with a flat network architecture. [Sri] How is this making the case of DMM ? We want the MN to access content= locally in the access network and we want localized routing. We have that = in the form of 6705 and 6909. The other approach is give localized IP addre= sses and have the content locally accessed. There are simply two many consi= derations and points may be valid, but when we bring all those assumptions.= We are bringing these points in a less logical manner without stating the = assumptions. Our response is similar to that in PS1. Also PS2 is referring to flat mobil= e network. PS3: Low scalability of centralized tunnel management and mobility context maintenance Setting up tunnels through a central anchor and maintaining mobility context for each MN usually requires more concentrated resources in a centralized design, thus reducing scalability. Distributing the tunnel maintenance function and the mobility context maintenance function among different network entities with proper signaling protocol design can increase scalability. PS4: Single point of failure and attack Centralized anchoring designs may be more vulnerable to single points of failures and attacks than a distributed system. The impact of a successful attack on a system with centralized mobility management can be far greater as well. PS5: Unnecessarily reserving resources to provide mobility support to nodes that do not need such support IP mobility support is not always required, and not every parameter of mobility context is always used. For example, some applications do not need a stable IP address during a handover to maintain session continuity. Sometimes, the entire application session runs while the terminal does not change the point of attachment. Besides, some sessions, e.g. SIP-based sessions, can handle mobility at the application layer and hence do not need IP mobility support; it is then more efficient to deactivate IP mobility support for such sessions. [Sri] Mobility systems today do support the aspect of service for the subs= criber, as "Simple IP", or "Mobile IP". A PDSN can assign a local IP addres= s and it does not have to be the home address. Network does have this intel= ligence, but what is missing is the client's ability to pick the correct ty= pe of IP address, among different IP addresses. The network is also the mis= sing the aspect of marking those addresses with proper properties. The curr= ently active drafts in IETF are to address this issue. We should talk about= these missing semantics. draft-bhandari-dhc-class-based-prefix-05 draft-korhonen-6man-prefix-properties-02 We need to find a suitable place to add these references. One place is to a= ppend to the end of Section 3 where it is talking about DMM work. The other= possibility is to add it as one of the considerations list in the Introduc= tion. Any suggestion? PS6: (Related problem) Mobility signaling overhead with peer-to-peer communication Wasting resources when mobility signaling (e.g., maintenance of the tunnel, keep alive signaling, etc.) is not turned off for peer-to-peer communication. Peer-to-peer communications have particular traffic patterns that often do not benefit from mobility support from the network. Thus, the associated mobility support signaling (e.g., maintenance of the tunnel, keep alive signaling, etc.) wastes network resources for no application gain. In such a case, it is better to enable mobility support selectively. [Sri] How is PS6 different from PS5 ? We talk about application's ability t= o pick the address with or without mobility properties in PS5. So, the traf= fic patterns can be localized based on the application requirements. But, e= ven if we take the argument that mobility is not required, operator needs v= isibility into these flows and so they can charge. Again, we have 6705 and= 6909 for adjusting to those traffic patterns. So, this P without those con= siderations is incomplete. Both PS5 and PS6 have been revised in version 08. If desired, we can clarif= y further (version 09?) that PS6 is about turning off the tunneling and sig= naling. PS7: (Related problem) Deployment with multiple mobility solutions There are already many variants and extensions of MIP. Deployment of new mobility management solutions can be challenging, and debugging difficult, when they must co-exist with solutions already in the field. PS8: Duplicate multicast traffic IP multicast distribution over architectures using IP mobility solutions (e.g. RFC6224) may lead to convergence of duplicated multicast subscriptions towards the downstream tunnel entity (e.g. MAG in PMIPv6). Concretely, when multicast subscription for individual mobile nodes is coupled with mobility tunnels (e.g. PMIPv6 tunnel), duplicate multicast subscription(s) is Chan (Ed.), et al. Expires February 3, 2014 [Page 10] Internet-Draft DMM-Reqs August 2013 prone to be received through different upstream paths. This problem may also exist or be more severe in a distributed mobility environment. [Sri] Is this for MN's from different LMA's attached to the same MAG ? It is one example. --_000_6E31144C030982429702B11D6746B98C370CBCD6szxeml557mbxchi_ Content-Type: text/html; charset="us-ascii" Content-Transfer-Encoding: quoted-printable

Sri,

 <= /p>

We have revised the probl= em statements in Section 4 in version 08. Our responses are in-line. Please= check again.

 <= /p>

4.  Problem Statement=

 

   The problems that can be addres= sed with DMM are summarized in the

   following:

 

   PS1:  Non-optimal routes

 

       &n= bsp; Routing via a centralized anchor often results in non-optimal

       &n= bsp; routes, thereby increasing the end-to-end delay.  The problem

       &n= bsp; is manifested, for example, when accessing a local server or

       &n= bsp; servers of a Content Delivery Network (CDN), or when receiving

       &n= bsp; locally available IP multicast or sending IP multicast packets.

 

   PS2:  Divergence from othe= r evolutionary trends in network

       &n= bsp; architectures such as distribution of content delivery.

 

       &n= bsp; Centralized mobility management can become non-optimal with a

       &n= bsp; flat network architecture.

 

   PS3:  Low scalability of c= entralized tunnel management and mobility

       &n= bsp; context maintenance

 

       &n= bsp; Setting up tunnels through a central anchor and maintaining=

        &= nbsp;mobility context for each MN usually requires more concentrated

       &n= bsp; resources in a centralized design, thus reducing scalability.

       &n= bsp; Distributing the tunnel maintenance function and the mobility

       &n= bsp; context maintenance function among different network entities

       &n= bsp; with proper signaling protocol design can increase scalability.

 

   PS4:  Single point of fail= ure and attack

 

       &n= bsp; Centralized anchoring designs may be more vulnerable to single

       &n= bsp; points of failures and attacks than a distributed system.  The

       &n= bsp; impact of a successful attack on a system with centralized<= /span>

       &n= bsp; mobility management can be far greater as well.

 

   PS5:  Unnecessary mobility= support to nodes that do not need it

 

       &n= bsp; IP mobility support is not always required, and not every

       &n= bsp; parameter of mobility context is always used.  For example,<= /o:p>

       &n= bsp; some applications do not need a stable IP address during a<= /span>

       &n= bsp; handover to maintain session continuity.  Sometimes, the entire

       &n= bsp; application session runs while the terminal does not change the

       &n= bsp; point of attachment.  Besides, some sessions, e.g.  SIP-base= d

       &n= bsp; sessions, can handle mobility at the application layer and<= /span>

       &n= bsp; hence do not need IP mobility support; it is then more

       &n= bsp; efficient to deactivate IP mobility support for such sessions.

 

   PS6:  (Related problem) Mo= bility signaling overhead with peer-to-peer

       &n= bsp; communication

 

       &n= bsp; Wasting resources when mobility signaling (e.g., maintenance of

       &n= bsp; the tunnel, keep alive signaling, etc.) is not turned off for

       &n= bsp; peer-to-peer communication.  Peer-to-peer communications have

       &n= bsp; particular traffic patterns that often do not benefit from<= /span>

       &n= bsp; mobility support from the network.  Thus, the associated

       &n= bsp; mobility support signaling (e.g., maintenance of the tunnel,

       &n= bsp; keep alive signaling, etc.) wastes network resources for no=

       &n= bsp; application gain.

 

   PS7:  (Related problem) De= ployment with multiple mobility solutions

 

       &n= bsp; There are already many variants and extensions of MIP.

       &n= bsp; Deployment of new mobility management solutions can be

       &n= bsp; challenging, and debugging difficult, when they must co-exist

       &n= bsp; with solutions already in the field.

 

   PS8:  Duplicate multicast = traffic

 

       &n= bsp; IP multicast distribution over architectures using IP mobility

       &n= bsp; solutions (e.g.  RFC6224) may lead to convergence of duplicated

       &n= bsp; multicast subscriptions towards the downstream tunnel entity

       &n= bsp; (e.g.  MAG in PMIPv6).  Concretely, when multicast subscript= ion

       &n= bsp; for individual mobile nodes is coupled with mobility tunnels

       &n= bsp; (e.g.  PMIPv6 tunnel), duplicate multicast subscription(s) is

       &n= bsp; prone to be received through different upstream paths.  This=

       &n= bsp; problem may also exist or be more severe in a distributed

       &n= bsp; mobility environment.

 <= /p>

The following have contri= buted to the above revisions in draft 08: Pierrick, Dapeng, and Jouni

 <= /p>

 <= /p>

From: dmm-bounces@ietf.org [mailto:dmm-bounces@ietf.org] On Behalf Of Sri Gundavelli (sgundave)
Sent: Sunday, August 25, 2013 10:04 PM
To: dmm@ietf.org
Subject: Re: [DMM] I-D Action: draft-ietf-dmm-requirements-07.txt

 

Please see inline for some comments.

 

Regards

Sri

 

 
 
4.  Problem St=
atement
 
   The pr=
oblems that can be addressed with DMM are summarized in the
   follow=
ing:
 
   PS1:&n=
bsp; Non-optimal routes
 
   &=
nbsp;     Routing via a centralized anchor often result=
s in a longer
   &=
nbsp;     route.  
[Sri] "longer route" ?&=
nbsp; I assume this is about routing/tx delay. Please re-word.
 
Accept and revised in=
 version 08
 
The problem is mani=
fested, for example, when accessing
   &=
nbsp;     a local server or servers of a Content Delive=
ry Network (CDN),
   &=
nbsp;     or when receiving locally available IP multic=
ast or sending IP
   &=
nbsp;     multicast packets.
 
[Sri] Does RFC 67=
05, or RFC 6909 does not address this issue ? May be the CDN example is inc=
orrect. 
 
CDN example is also a=
n important driver: distributing the mobility anchor is more compatible wit=
h distributing and replicating the contents, which may be more general to t=
he particular protocol solutions. It may also include the case that there i=
s a CDN server is in a neighboring network. Suggestions to reword (for vers=
ion 09) are welcome.  
 
   PS2:&n=
bsp; Divergence from other evolutionary trends in network=

   &=
nbsp;     architectures such as distribution of content=
 delivery.
 
   &=
nbsp;     Centralized mobility management can become no=
n-optimal with a
   &=
nbsp;     flat network architecture.<=
/pre>

 


[Sri] How is this=
 making the case of DMM ? We want the MN to access content locally in the a=
ccess network and we want localized routing. We have that in the form of 67=
05 and 6909. The other approach is give localized IP addresses and have the=
 content locally accessed. There are simply two many considerations and poi=
nts may be valid, but when we bring all those assumptions. We are bringing =
these points in a less logical manner without stating the assumptions.


 =



Our response is simi=
lar to that in PS1. Also PS2 is referring to flat mobile network. 


 


   PS3:&n=
bsp; Low scalability of centralized tunnel management and mobility


   &=
nbsp;     context maintenance


 


   &=
nbsp;     Setting up tunnels through a central anchor a=
nd maintaining


   &=
nbsp;     mobility context for each MN usually requires=
 more concentrated


   &=
nbsp;     resources in a centralized design, thus reduc=
ing scalability.


   &=
nbsp;     Distributing the tunnel maintenance function =
and the mobility


   &=
nbsp;     context maintenance function among different =
network entities


   &=
nbsp;     with proper signaling protocol design can inc=
rease scalability.


 


   PS4:&n=
bsp; Single point of failure and attack


 


   &=
nbsp;     Centralized anchoring designs may be more vul=
nerable to single


   &=
nbsp;     points of failures and attacks than a distrib=
uted system.  The


   &=
nbsp;     impact of a successful attack on a system wit=
h centralized


   &=
nbsp;     mobility management can be far greater as wel=
l.


 


   PS5:&n=
bsp; Unnecessarily reserving resources to provide mobility support


   &=
nbsp;     to nodes that do not need such support


 


   &=
nbsp;     IP mobility support is not always required, a=
nd not every


   &=
nbsp;     parameter of mobility context is always used.=
  For example,


   &=
nbsp;     some applications do not need a stable IP add=
ress during a


   &=
nbsp;     handover to maintain session continuity. =
; Sometimes, the entire


   &=
nbsp;     application session runs while the terminal d=
oes not change the


   &=
nbsp;     point of attachment.  Besides, some sess=
ions, e.g.  SIP-based


   &=
nbsp;     sessions, can handle mobility at the applicat=
ion layer and


   &=
nbsp;     hence do not need IP mobility support; it is =
then more


   &=
nbsp;     efficient to deactivate IP mobility support f=
or such sessions.


 


[Sri]  Mobility systems toda=
y do support the aspect of service for the subscriber, as "Simple IP&q=
uot;, or "Mobile IP". A PDSN can assign a local IP address and it=
 does not have to be the home address. Network does have this intelligence,=
 but what is missing is the client's ability to pick the correct type of IP=
 address, among different IP addresses. The network is also the missing the=
 aspect of marking those addresses with proper properties. The currently ac=
tive drafts in IETF are to address this issue. We should talk about these m=
issing semantics.<=
/pre>

draft-bhandari-dhc-class-based-prefix-05  


draft-korhonen-6man-prefix-properties-02<=
/span>


 


We need to find a sui=
table place to add these references. One place is to append to the end of S=
ection 3 where it is talking about DMM work. The other possibility is to ad=
d it as one of the considerations list in the Introduction. Any suggestion?=



   PS6:&n=
bsp; (Related problem) Mobility signaling overhead with peer-to-peer


   &=
nbsp;     communication


 


   &=
nbsp;     Wasting resources when mobility signaling (e.=
g., maintenance of


   &=
nbsp;     the tunnel, keep alive signaling, etc.) is no=
t turned off for


   &=
nbsp;     peer-to-peer communication.  Peer-to-pee=
r communications have


   &=
nbsp;     particular traffic patterns that often do not=
 benefit from


   &=
nbsp;     mobility support from the network.  Thus=
, the associated


   &=
nbsp;     mobility support signaling (e.g., maintenance=
 of the tunnel,


   &=
nbsp;     keep alive signaling, etc.) wastes network re=
sources for no


   &=
nbsp;     application gain.  In such a case, it is=
 better to enable


   &=
nbsp;     mobility support selectively.


 


[Sri] How is PS6 different from P=
S5 ? We talk about application's ability to pick the address with or withou=
t mobility properties in PS5. So, the traffic patterns can be localized bas=
ed on the application requirements. But, even if we take the argument that =
mobility is not required, operator needs visibility into these flows and so=
 they can charge.  Again, we have 6705 and 6909 for adjusting to those=
 traffic patterns. So, this P without those considerations is incomplete.


 


Both PS5 and PS6 have=
 been revised in version 08. If desired, we can clarify further (version 09=
?) that PS6 is about turning off the tunneling and signaling. 


   PS7:&n=
bsp; (Related problem) Deployment with multiple mobility solutions


 


   &=
nbsp;     There are already many variants and extension=
s of MIP.


    =
     Deployment of new mobility management solutio=
ns can be


   &=
nbsp;     challenging, and debugging difficult, when th=
ey must co-exist


   &=
nbsp;     with solutions already in the field.


 


 


 


   PS8:  Duplicate m=
ulticast traffic


 


      =
;   IP multicast distribution over architectures using IP mobilit=
y


      =
;   solutions (e.g.  RFC6224) may lead to convergence of dup=
licated


      =
;   multicast subscriptions towards the downstream tunnel entity<=
o:p>


      =
;   (e.g.  MAG in PMIPv6).  Concretely, when multicast =
subscription


      &nbs=
p;  for individual mobile nodes is coupled with mobility tunnels<=
o:p>


      =
;   (e.g.  PMIPv6 tunnel), duplicate multicast subscription(=
s) is


 


 


 


Chan (Ed.), et al.   =
;   Expires February 3, 2014      &=
nbsp;        [Page 10]=



 


Internet-Draft   &nb=
sp;            =
  DMM-Reqs          =
           August 2013


 


 


      =
;   prone to be received through different upstream paths.  =
This


      =
;   problem may also exist or be more severe in a distributed


      =
;   mobility environment.


 


[Sri] Is this for MN's from diffe=
rent LMA's attached to the same MAG ?


 

It is one example.

--_000_6E31144C030982429702B11D6746B98C370CBCD6szxeml557mbxchi_-- From h.anthony.chan@huawei.com Thu Sep 26 12:11:26 2013 Return-Path: X-Original-To: dmm@ietfa.amsl.com Delivered-To: dmm@ietfa.amsl.com Received: from localhost (localhost [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id A432121F9E54 for ; Thu, 26 Sep 2013 12:11:24 -0700 (PDT) X-Virus-Scanned: amavisd-new at amsl.com X-Spam-Flag: NO X-Spam-Score: -6.519 X-Spam-Level: X-Spam-Status: No, score=-6.519 tagged_above=-999 required=5 tests=[AWL=0.079, BAYES_00=-2.599, HTML_MESSAGE=0.001, RCVD_IN_DNSWL_MED=-4] Received: from mail.ietf.org ([12.22.58.30]) by localhost (ietfa.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id F8QKp+UXj8Lq for ; Thu, 26 Sep 2013 12:11:16 -0700 (PDT) Received: from lhrrgout.huawei.com (lhrrgout.huawei.com [194.213.3.17]) by ietfa.amsl.com (Postfix) with ESMTP id 06B5221F9C87 for ; Thu, 26 Sep 2013 12:11:13 -0700 (PDT) Received: from 172.18.7.190 (EHLO lhreml204-edg.china.huawei.com) ([172.18.7.190]) by lhrrg02-dlp.huawei.com (MOS 4.3.5-GA FastPath queued) with ESMTP id AVX00230; Thu, 26 Sep 2013 19:11:12 +0000 (GMT) Received: from LHREML401-HUB.china.huawei.com (10.201.5.240) by lhreml204-edg.china.huawei.com (172.18.7.223) with Microsoft SMTP Server (TLS) id 14.3.146.0; Thu, 26 Sep 2013 20:10:18 +0100 Received: from SZXEML413-HUB.china.huawei.com (10.82.67.152) by lhreml401-hub.china.huawei.com (10.201.5.240) with Microsoft SMTP Server (TLS) id 14.3.146.0; Thu, 26 Sep 2013 20:11:08 +0100 Received: from szxeml557-mbx.china.huawei.com ([169.254.5.217]) by szxeml413-hub.china.huawei.com ([10.82.67.152]) with mapi id 14.03.0146.000; Fri, 27 Sep 2013 03:11:02 +0800 From: h chan To: "Sri Gundavelli (sgundave)" , "dmm@ietf.org" Thread-Topic: [DMM] I-D Action: draft-ietf-dmm-requirements-07.txt Thread-Index: AQHOogjoa3/L9muNGkS5wozrKtEQEpnXRZkQgABo14CAAOMQoA== Date: Thu, 26 Sep 2013 19:11:01 +0000 Message-ID: <6E31144C030982429702B11D6746B98C370CBD2D@szxeml557-mbx.china.huawei.com> References: <1377486056.79906.YahooMailNeo@web163805.mail.gq1.yahoo.com> <24C0F3E22276D9438D6F366EB89FAEA8116A4D4F@xmb-aln-x03.cisco.com> Accept-Language: en-US, zh-CN Content-Language: en-US X-MS-Has-Attach: X-MS-TNEF-Correlator: x-originating-ip: [10.192.11.128] Content-Type: multipart/alternative; boundary="_000_6E31144C030982429702B11D6746B98C370CBD2Dszxeml557mbxchi_" MIME-Version: 1.0 X-CFilter-Loop: Reflected Subject: Re: [DMM] I-D Action: draft-ietf-dmm-requirements-07.txt X-BeenThere: dmm@ietf.org X-Mailman-Version: 2.1.12 Precedence: list List-Id: Distributed Mobility Management Working Group List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Thu, 26 Sep 2013 19:11:26 -0000 --_000_6E31144C030982429702B11D6746B98C370CBD2Dszxeml557mbxchi_ Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: quoted-printable Sri, Regarding your comments on the requirements, the motivation of REQ2 has bee= n revised in version 08. The responses to your comments are in-line. Pleas= e check. 5.2. Transparency to Upper Layers when needed REQ2: Transparency to Upper Layers when needed DMM solutions MUST provide transparent mobility support above the IP layer when needed. Such transparency is needed, for example, when, upon change of point of attachment to the network, an application flow cannot cope with a change in the IP address. However, it is not always necessary to maintain a stable home IP address or prefix for every application or at all times for a mobile node. Motivation: The motivation of this requirement is to enable more efficient routing and more efficient use of network resources by selecting an IP address or prefix according to whether mobility support is needed and by not maintaining context at the mobility anchor when there is no such need. This requirement addresses the problem PS5 as well as the related problem PS6 stated in Section 4. The following have contributed to the revisions in draft 08: Pierrick, Dape= ng, and Jouni From: dmm-bounces@ietf.org [mailto:dmm-bounces= @ietf.org] On Behalf Of Sri Gundavelli (sgundave) Sent: Sunday, August 25, 2013 10:04 PM To: dmm@ietf.org Subject: Re: [DMM] I-D Action: draft-ietf-dmm-requirements-07.txt Please see inline for some comments. 5. Requirements After comparing distributed mobility management against centralized deployment in Section 3, this section identifies the following requirements: 5.1. Distributed processing REQ1: Distributed processing IP mobility, network access and routing solutions provided by DMM MUST enable distributed processing for mobility management so that traffic does not need to traverse centrally deployed mobility anchors and thereby avoid non-optimal routes. Motivation: This requirement is motivated by current trends in network evolution: (a) it is cost- and resource-effective to cache and distribute content by combining distributed mobility anchors with caching systems (e.g., CDN); (b) the significantly larger number of mobile nodes and flows call for improved scalability; (c) single points of failure are avoided in a distributed system; (d) threats against centrally deployed anchors, e.g., home agent and local mobility anchor, are mitigated in a distributed system. This requirement addresses the problems PS1, PS2, PS3, and PS4 described in Section 4. (Existing route optimization is only a host- based solution. On the other hand, localized routing with PMIPv6 addresses only a part of the problem where both the MN and the CN are located in the PMIP domain and attached to a MAG, and is not applicable when the CN is outside the PMIP domain.) [Sri] I'm still stuck on the CDN example driving this requirement. The requirements are general. In addition, whether specific solutions are s= ufficient is the role of the separate draft on gap analysis 5.2. Transparency to Upper Layers when needed REQ2: Transparency to Upper Layers when needed DMM solutions MUST provide transparent mobility support above the IP layer when needed. Such transparency is needed, for example, when, upon change of point of attachment to the network, an application flow cannot cope with a change in the IP address. However, it is not always necessary to maintain a stable home IP address or prefix for every application or at all times for a mobile node. [Sri] Please reflect the two key aspects of this requirement: * Network can assign IP addresses with different properties; It carr= ies those properties * Applications have different requirements and will pick the address= with the correct property. Motivation: The motivation of this requirement is to enable more efficient use of network resources and more efficient routing by not maintaining context at the mobility anchor when there is no such need. This requirement addresses the problem PS5 as well as the related problem PS6 stated in Section 4. --_000_6E31144C030982429702B11D6746B98C370CBD2Dszxeml557mbxchi_ Content-Type: text/html; charset="us-ascii" Content-Transfer-Encoding: quoted-printable

Sri,

 <= /p>

Regarding your comments on the requirements, the motivation of R= EQ2 has been revised in version 08. The responses to your comments are in-l= ine.  Please check.

 <= /p>

5.2.  Transparency to Upper Layers when= needed

 

   REQ2:  Transparency to Upp= er Layers when needed

 

       &n= bsp;  DMM solutions MUST provide transparent mobility support above

       &n= bsp;  the IP layer when needed.  Such transparency is needed, for=

       &n= bsp;  example, when, upon change of point of attachment to the

       &n= bsp;  network, an application flow cannot cope with a change in the

       &n= bsp;  IP address.  However, it is not always necessary to maintai= n a

       &n= bsp;  stable home IP address or prefix for every application or at

        &= nbsp; all times for a mobile node.

 

       &n= bsp;  Motivation: The motivation of this requirement is to enable=

       &n= bsp;  more efficient routing and more efficient use of network

       &n= bsp;  resources by selecting an IP address or prefix according to

        =   whether mobility support is needed and by not maintai= ning

       &n= bsp;  context at the mobility anchor when there is no such need.<= /o:p>

 

   This requirement addresses the = problem PS5 as well as the related

   problem PS6 stated in Section 4= .

 <= /p>

The following have contri= buted to the revisions in draft 08: Pierrick, Dapeng, and Jouni<= /span>

 <= /p>

 <= /p>

From: dmm-bounces@ietf.org [mailto:dmm-bounces@ietf.org] On Behalf Of Sri Gundavelli (sgundave)
Sent: Sunday, August 25, 2013 10:04 PM
To: dmm@ietf.org
Subject: Re: [DMM] I-D Action: draft-ietf-dmm-requirements-07.txt

 

Please see inline for some comments.

 

5.  Requiremen=
ts
 
   After =
comparing distributed mobility management against centralized
   deploy=
ment in Section 3, this section identifies the following<=
/pre>

   requir=
ements:


 


5.1.  Distribu=
ted processing


 


   REQ1:&=
nbsp; Distributed processing


 


   &=
nbsp;      IP mobility, network access and routing=
 solutions provided by


   &=
nbsp;      DMM MUST enable distributed processing =
for mobility management


   &=
nbsp;      so that traffic does not need to traver=
se centrally deployed


   &=
nbsp;      mobility anchors and thereby avoid non-=
optimal routes.


 


   &=
nbsp;      Motivation: This requirement is motivat=
ed by current trends in


   &=
nbsp;      network evolution: (a) it is cost- and =
resource-effective to


   &=
nbsp;      cache and distribute content by combini=
ng distributed mobility


   &=
nbsp;      anchors with caching systems (e.g., CDN=
); (b) the


   &=
nbsp;      significantly larger number of mobile n=
odes and flows call for


   &=
nbsp;      improved scalability; (c) single points=
 of failure are avoided


   &=
nbsp;      in a distributed system; (d) threats ag=
ainst centrally


   &=
nbsp;      deployed anchors, e.g., home agent and =
local mobility anchor,


   &=
nbsp;      are mitigated in a distributed system.<=
o:p>


 


   This r=
equirement addresses the problems PS1, PS2, PS3, and PS4<=
/pre>

   descri=
bed in Section 4.  (Existing route optimization is only a host-


   based =
solution.  On the other hand, localized routing with PMIPv6=



   addres=
ses only a part of the problem where both the MN and the CN are<=
/span>


   locate=
d in the PMIP domain and attached to a MAG, and is not

   applic=
able when the CN is outside the PMIP domain.)


 


[Sri] I'm still stuck on the CDN =
example driving this requirement.


 


The requirements are =
general. In addition, whether specific solutions are sufficient is the role=
 of the separate draft on gap analysis


 


5.2.  Transpar=
ency to Upper Layers when needed


 


   REQ2:&=
nbsp; Transparency to Upper Layers when needed


 


   &=
nbsp;      DMM solutions MUST provide transparent =
mobility support above


   &=
nbsp;      the IP layer when needed.  Such tr=
ansparency is needed, for


   &=
nbsp;      example, when, upon change of point of =
attachment to the


   &=
nbsp;      network, an application flow cannot cop=
e with a change in the


   &=
nbsp;      IP address.  However, it is not al=
ways necessary to maintain a


   &=
nbsp;      stable home IP address or prefix for ev=
ery application or at


   &=
nbsp;      all times for a mobile node.=



 


 


[Sri] Please refl=
ect the two key aspects of this requirement:


·        Network can assign I=
P addresses with different properties; It carries those properties


·        Applications have di=
fferent requirements and will pick the address with the correct property.


 


   &=
nbsp;      Motivation: The motivation of this requ=
irement is to enable


   &=
nbsp;      more efficient use of network resources=
 and more efficient


   &=
nbsp;      routing by not maintaining context at t=
he mobility anchor when


   &=
nbsp;      there is no such need.


   This r=
equirement addresses the problem PS5 as well as the related


   proble=
m PS6 stated in Section 4.


 

 

--_000_6E31144C030982429702B11D6746B98C370CBD2Dszxeml557mbxchi_-- From internet-drafts@ietf.org Sat Sep 28 08:03:36 2013 Return-Path: X-Original-To: dmm@ietfa.amsl.com Delivered-To: dmm@ietfa.amsl.com Received: from localhost (localhost [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id E050511E8142; Sat, 28 Sep 2013 08:03:36 -0700 (PDT) X-Virus-Scanned: amavisd-new at amsl.com X-Spam-Flag: NO X-Spam-Score: -102.569 X-Spam-Level: X-Spam-Status: No, score=-102.569 tagged_above=-999 required=5 tests=[AWL=0.031, BAYES_00=-2.599, NO_RELAYS=-0.001, USER_IN_WHITELIST=-100] Received: from mail.ietf.org ([12.22.58.30]) by localhost (ietfa.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id Kbq3rjXDqoix; Sat, 28 Sep 2013 08:03:35 -0700 (PDT) Received: from ietfa.amsl.com (localhost [IPv6:::1]) by ietfa.amsl.com (Postfix) with ESMTP id D7C6111E80E2; Sat, 28 Sep 2013 08:03:06 -0700 (PDT) MIME-Version: 1.0 Content-Type: text/plain; charset="utf-8" Content-Transfer-Encoding: quoted-printable From: internet-drafts@ietf.org To: i-d-announce@ietf.org X-Test-IDTracker: no X-IETF-IDTracker: 4.72 Auto-Submitted: auto-generated Precedence: bulk Message-ID: <20130928150306.9951.52677.idtracker@ietfa.amsl.com> Date: Sat, 28 Sep 2013 08:03:06 -0700 Cc: dmm@ietf.org Subject: [DMM] I-D Action: draft-ietf-dmm-requirements-09.txt X-BeenThere: dmm@ietf.org X-Mailman-Version: 2.1.12 List-Id: Distributed Mobility Management Working Group List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Sat, 28 Sep 2013 15:03:37 -0000 A New Internet-Draft is available from the on-line Internet-Drafts director= ies. This draft is a work item of the Distributed Mobility Management Working G= roup of the IETF. Title : Requirements for Distributed Mobility Management Author(s) : H Anthony Chan Dapeng Liu Pierrick Seite Hidetoshi Yokota Jouni Korhonen Filename : draft-ietf-dmm-requirements-09.txt Pages : 20 Date : 2013-09-28 Abstract: This document defines the requirements for Distributed Mobility Management (DMM). The hierarchical structure in traditional wireless networks has led primarily to centralized deployment models. As some wireless networks are evolving away from the hierarchical structure, such as in moving the content delivery servers closer to the users, a distributed model for mobility management can be useful to them. The IETF datatracker status page for this draft is: https://datatracker.ietf.org/doc/draft-ietf-dmm-requirements There's also a htmlized version available at: http://tools.ietf.org/html/draft-ietf-dmm-requirements-09 A diff from the previous version is available at: http://www.ietf.org/rfcdiff?url2=3Ddraft-ietf-dmm-requirements-09 Please note that it may take a couple of minutes from the time of submission until the htmlized version and diff are available at tools.ietf.org. Internet-Drafts are also available by anonymous FTP at: ftp://ftp.ietf.org/internet-drafts/ From h.anthony.chan@huawei.com Sat Sep 28 08:26:02 2013 Return-Path: X-Original-To: dmm@ietfa.amsl.com Delivered-To: dmm@ietfa.amsl.com Received: from localhost (localhost [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id 4D24211E8139 for ; Sat, 28 Sep 2013 08:26:02 -0700 (PDT) X-Virus-Scanned: amavisd-new at amsl.com X-Spam-Flag: NO X-Spam-Score: -6.535 X-Spam-Level: X-Spam-Status: No, score=-6.535 tagged_above=-999 required=5 tests=[AWL=0.063, BAYES_00=-2.599, HTML_MESSAGE=0.001, RCVD_IN_DNSWL_MED=-4] Received: from mail.ietf.org ([12.22.58.30]) by localhost (ietfa.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id nIJmRTBXHrdE for ; Sat, 28 Sep 2013 08:25:48 -0700 (PDT) Received: from lhrrgout.huawei.com (lhrrgout.huawei.com [194.213.3.17]) by ietfa.amsl.com (Postfix) with ESMTP id 5F94421E8051 for ; Sat, 28 Sep 2013 08:25:45 -0700 (PDT) Received: from 172.18.7.190 (EHLO lhreml204-edg.china.huawei.com) ([172.18.7.190]) by lhrrg01-dlp.huawei.com (MOS 4.3.7-GA FastPath queued) with ESMTP id AYJ38902; Sat, 28 Sep 2013 15:25:40 +0000 (GMT) Received: from LHREML406-HUB.china.huawei.com (10.201.5.243) by lhreml204-edg.china.huawei.com (172.18.7.223) with Microsoft SMTP Server (TLS) id 14.3.146.0; Sat, 28 Sep 2013 16:24:41 +0100 Received: from SZXEML413-HUB.china.huawei.com (10.82.67.152) by lhreml406-hub.china.huawei.com (10.201.5.243) with Microsoft SMTP Server (TLS) id 14.3.146.0; Sat, 28 Sep 2013 16:25:35 +0100 Received: from szxeml557-mbx.china.huawei.com ([169.254.5.217]) by szxeml413-hub.china.huawei.com ([10.82.67.152]) with mapi id 14.03.0146.000; Sat, 28 Sep 2013 23:25:29 +0800 From: h chan To: "Sri Gundavelli (sgundave)" , "dmm@ietf.org" Thread-Topic: [DMM] I-D Action: draft-ietf-dmm-requirements-07.txt Thread-Index: AQHOogjoa3/L9muNGkS5wozrKtEQEpnXRZkQgABo14CAAJoKEIADLS4w Date: Sat, 28 Sep 2013 15:25:29 +0000 Message-ID: <6E31144C030982429702B11D6746B98C370CBF59@szxeml557-mbx.china.huawei.com> References: <1377486056.79906.YahooMailNeo@web163805.mail.gq1.yahoo.com> <24C0F3E22276D9438D6F366EB89FAEA8116A4D4F@xmb-aln-x03.cisco.com> Accept-Language: en-US, zh-CN Content-Language: en-US X-MS-Has-Attach: X-MS-TNEF-Correlator: x-originating-ip: [10.46.64.17] Content-Type: multipart/alternative; boundary="_000_6E31144C030982429702B11D6746B98C370CBF59szxeml557mbxchi_" MIME-Version: 1.0 X-CFilter-Loop: Reflected Subject: Re: [DMM] I-D Action: draft-ietf-dmm-requirements-07.txt X-BeenThere: dmm@ietf.org X-Mailman-Version: 2.1.12 Precedence: list List-Id: Distributed Mobility Management Working Group List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Sat, 28 Sep 2013 15:26:02 -0000 --_000_6E31144C030982429702B11D6746B98C370CBF59szxeml557mbxchi_ Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: quoted-printable Sri, We have revised the Introduction Section and PS1 in version 09. Our respons= es are in-line. The following have contributed to the above revisions in draft 09: Pierrick= , Dapeng, and Jouni From: dmm-bounces@ietf.org [mailto:dmm-bounces= @ietf.org] On Behalf Of Sri Gundavelli (sgundave) Sent: Sunday, August 25, 2013 10:04 PM To: dmm@ietf.org Subject: Re: [DMM] I-D Action: draft-ietf-dmm-requirements-07.txt Please see inline for some comments. Regards Sri ....... 4. Problem Statement The problems that can be addressed with DMM are summarized in the following: PS1: Non-optimal routes Routing via a centralized anchor often results in a longer route. [Sri] "longer route" ? I assume this is about routing/tx delay. Please re-= word. Accept and revised in version 08 The problem is manifested, for example, when accessing a local server or servers of a Content Delivery Network (CDN), or when receiving locally available IP multicast or sending IP multicast packets. [Sri] Does RFC 6705, or RFC 6909 does not address this issue ? May be the C= DN example is incorrect. RFC6909 is clarified further in the Introduction Section in version 09. (1) Mobile users are, more than ever, consuming Internet content; such traffic imposes new requirements on mobile core networks for data traffic delivery. The presence of content providers closer to Internet Service Providers (ISP) network requires taking into account local Content Delivery Networks (CDNs) while providing mobility services. Moreover, when the traffic demand exceeds available capacity, service providers need to implement new strategies such as selective IPv4 traffic offload (e.g. [RFC6909], 3GPP work items LIPA/SIPTO [TS.23.401]) through alternative access networks (e.g. WLAN) [Paper- Mobile.Data.Offloading]. RFC6705 is clarified under PS1 in version 09 PS1: Non-optimal routes Routing via a centralized anchor often results in non-optimal routes, thereby increasing the end-to-end delay. The problem is manifested, for example, when accessing a nearby server or servers of a Content Delivery Network (CDN), or when receiving locally available IP multicast or sending IP multicast packets. (Existing route optimization is only a host-based solution. On the other hand, localized routing with PMIPv6 [RFC6705] addresses only a part of the problem where both the MN and the CN are located in the PMIP domain and attached to a MAG, and is not applicable when the CN is outside the PMIP domain or does not behave like an MN.) PS2: Divergence from other evolutionary trends in network architectures such as distribution of content delivery. Centralized mobility management can become non-optimal with a flat network architecture. [Sri] How is this making the case of DMM ? We want the MN to access content= locally in the access network and we want localized routing. We have that = in the form of 6705 and 6909. The other approach is give localized IP addre= sses and have the content locally accessed. There are simply two many consi= derations and points may be valid, but when we bring all those assumptions.= We are bringing these points in a less logical manner without stating the = assumptions. Our response is similar to that in PS1. Also PS2 is referring to flat mobil= e network. PS3: Low scalability of centralized tunnel management and mobility context maintenance Setting up tunnels through a central anchor and maintaining mobility context for each MN usually requires more concentrated resources in a centralized design, thus reducing scalability. Distributing the tunnel maintenance function and the mobility context maintenance function among different network entities with proper signaling protocol design can increase scalability. PS4: Single point of failure and attack Centralized anchoring designs may be more vulnerable to single points of failures and attacks than a distributed system. The impact of a successful attack on a system with centralized mobility management can be far greater as well. PS5: Unnecessarily reserving resources to provide mobility support to nodes that do not need such support IP mobility support is not always required, and not every parameter of mobility context is always used. For example, some applications do not need a stable IP address during a handover to maintain session continuity. Sometimes, the entire application session runs while the terminal does not change the point of attachment. Besides, some sessions, e.g. SIP-based sessions, can handle mobility at the application layer and hence do not need IP mobility support; it is then more efficient to deactivate IP mobility support for such sessions. [Sri] Mobility systems today do support the aspect of service for the subs= criber, as "Simple IP", or "Mobile IP". A PDSN can assign a local IP addres= s and it does not have to be the home address. Network does have this intel= ligence, but what is missing is the client's ability to pick the correct ty= pe of IP address, among different IP addresses. The network is also the mis= sing the aspect of marking those addresses with proper properties. The curr= ently active drafts in IETF are to address this issue. We should talk about= these missing semantics. draft-bhandari-dhc-class-based-prefix-05 draft-korhonen-6man-prefix-properties-02 These references have been added in the Introduction Section in version 09 (2) Today's mobile networks present service providers with new challenges. Mobility patterns indicate that mobile nodes often remain attached to the same point of attachment for considerable periods of time [Paper-Locating.User]. Specific IP mobility management support is not required for applications that launch and complete their sessions while the mobile node is connected to the same point of attachment. However, currently, IP mobility support is designed for always-on operation, maintaining all parameters of the context for each mobile subscriber for as long as they are connected to the network. This can result in a waste of resources and unnecessary costs for the service provider. Infrequent node mobility coupled with application intelligence suggest that mobility support could be provided selectively such as in [I-D.bhandari-dhc-class-based- prefix] and [I-D.korhonen-6man-prefix-properties], thus reducing the amount of context maintained in the network. --_000_6E31144C030982429702B11D6746B98C370CBF59szxeml557mbxchi_ Content-Type: text/html; charset="us-ascii" Content-Transfer-Encoding: quoted-printable

Sri,

 <= /p>

We have revised the Introduction Section and PS1 in version 0= 9. Our responses are in-line.

 <= /p>

The following have contri= buted to the above revisions in draft 09<= /span>: Pierrick, Dapeng, and Jouni

 <= /p>

 <= /p>

From: dmm-bounces@ietf.org [mailto:dmm-bounces@ietf.org] On Behalf Of Sri Gundavelli (sgundave)
Sent: Sunday, August 25, 2013 10:04 PM
To: dmm@ietf.org
Subject: Re: [DMM] I-D Action: draft-ietf-dmm-requirements-07.txt

 

Please see inline for some comments.

 

Regards

Sri

 

…….
 
4.  Problem St=
atement
 
   The pr=
oblems that can be addressed with DMM are summarized in the
   follow=
ing:
 
   PS1:&n=
bsp; Non-optimal routes
 
   &=
nbsp;     Routing via a centralized anchor often result=
s in a longer
   &=
nbsp;     route.  
[Sri] "longer route" ?&=
nbsp; I assume this is about routing/tx delay. Please re-word.
 
Accept and revised in=
 version 08
 
The problem is mani=
fested, for example, when accessing
   &=
nbsp;     a local server or servers of a Content Delive=
ry Network (CDN),
   &=
nbsp;     or when receiving locally available IP multic=
ast or sending IP
   &=
nbsp;     multicast packets.
 
[Sri] Does RFC 67=
05, or RFC 6909 does not address this issue ? May be the CDN example is inc=
orrect. 
 
RFC6909 is clarified =
further in the Introduction Section in version 09.  =

   (1)  Mobile users are, more than ever, c= onsuming Internet content;

        such traffic im= poses new requirements on mobile core networks

        for data traffi= c delivery.  The presence of content providers

        closer to Inter= net Service Providers (ISP) network requires

        taking into acc= ount local Content Delivery Networks (CDNs) while

        providing mobil= ity services.  Moreover, when the traffic demand

        exceeds availab= le capacity, service providers need to implement

        new strategies = such as selective IPv4 traffic offload (e.g.

        [RFC6909], 3GPP= work items LIPA/SIPTO [TS.23.401]) through

        alternative acc= ess networks (e.g.  WLAN) [Paper-

        =
Mobile.Data.Offloading].  
 =

RFC6705 is clarified=
 under PS1 in version 09

   PS1:  Non-optimal routes

 

         Routing v= ia a centralized anchor often results in non-optimal

         routes, t= hereby increasing the end-to-end delay.  The problem=

         is manife= sted, for example, when accessing a nearby server or

         servers o= f a Content Delivery Network (CDN), or when receiving

         locally a= vailable IP multicast or sending IP multicast packets.

        = (Existing route optimization is only a host-based solution.  On<= /o:p>

        = the other hand, localized routing with PMIPv6 [RFC6705]<= /p>

        = addresses only a part of the problem where both the MN and the<= /span>

        = CN are located in the PMIP domain and attached to a MAG, and is=

        = not applicable when the CN is outside the PMIP domain or does

        = not behave like an MN.)

 
 
 
   PS2:&n=
bsp; Divergence from other evolutionary trends in network=

   &=
nbsp;     architectures such as distribution of content=
 delivery.
 
   &=
nbsp;     Centralized mobility management can become no=
n-optimal with a
   &=
nbsp;     flat network architecture.<=
/pre>

 


[Sri] How is this=
 making the case of DMM ? We want the MN to access content locally in the a=
ccess network and we want localized routing. We have that in the form of 67=
05 and 6909. The other approach is give localized IP addresses and have the=
 content locally accessed. There are simply two many considerations and poi=
nts may be valid, but when we bring all those assumptions. We are bringing =
these points in a less logical manner without stating the assumptions.


 =



Our response is simi=
lar to that in PS1. Also PS2 is referring to flat mobile network. 


 


   PS3:&n=
bsp; Low scalability of centralized tunnel management and mobility


   &=
nbsp;     context maintenance


 


   &=
nbsp;     Setting up tunnels through a central anchor a=
nd maintaining


   &=
nbsp;     mobility context for each MN usually requires=
 more concentrated


   &=
nbsp;     resources in a centralized design, thus reduc=
ing scalability.


   &=
nbsp;     Distributing the tunnel maintenance function =
and the mobility


   &=
nbsp;     context maintenance function among different =
network entities


   &=
nbsp;     with proper signaling protocol design can inc=
rease scalability.


 


   PS4:&n=
bsp; Single point of failure and attack


 


   &=
nbsp;     Centralized anchoring designs may be more vul=
nerable to single


   &=
nbsp;     points of failures and attacks than a distrib=
uted system.  The


   &=
nbsp;     impact of a successful attack on a system wit=
h centralized


   &=
nbsp;     mobility management can be far greater as wel=
l.


 


   PS5:&n=
bsp; Unnecessarily reserving resources to provide mobility support


   &=
nbsp;     to nodes that do not need such support


 


   &=
nbsp;     IP mobility support is not always required, a=
nd not every


   &=
nbsp;     parameter of mobility context is always used.=
  For example,


   &=
nbsp;     some applications do not need a stable IP add=
ress during a


   &=
nbsp;     handover to maintain session continuity. =
; Sometimes, the entire


   &=
nbsp;     application session runs while the terminal d=
oes not change the


   &=
nbsp;     point of attachment.  Besides, some sess=
ions, e.g.  SIP-based


   &=
nbsp;     sessions, can handle mobility at the applicat=
ion layer and


   &=
nbsp;     hence do not need IP mobility support; it is =
then more


   &=
nbsp;     efficient to deactivate IP mobility support f=
or such sessions.


 


[Sri]  Mobility systems toda=
y do support the aspect of service for the subscriber, as "Simple IP&q=
uot;, or "Mobile IP". A PDSN can assign a local IP address and it=
 does not have to be the home address. Network does have this intelligence,=
 but what is missing is the client's ability to pick the correct type of IP=
 address, among different IP addresses. The network is also the missing the=
 aspect of marking those addresses with proper properties. The currently ac=
tive drafts in IETF are to address this issue. We should talk about these m=
issing semantics.<=
/pre>

draft-bhandari-dhc-class-based-prefix-05  


draft-korhonen-6man-prefix-properties-02<=
/span>


 


These references have=
 been added in the Introduction Section in version 09

 


   (2)  Today's mobile networks present ser=
vice providers with new


        challenges.&nbs=
p; Mobility patterns indicate that mobile nodes often


        remain attached=
 to the same point of attachment for considerable


        periods of time=
 [Paper-Locating.User].  Specific IP mobility


        management supp=
ort is not required for applications that launch


        and complete th=
eir sessions while the mobile node is connected


        to the same poi=
nt of attachment.  However, currently, IP


        mobility suppor=
t is designed for always-on operation,


        maintaining all=
 parameters of the context for each mobile


        subscriber for =
as long as they are connected to the network.


        This can result=
 in a waste of resources and unnecessary costs


        for the service=
 provider.  Infrequent node mobility coupled with

        application int=
elligence suggest that mobility support could be


        provided select=
ively
such as in [I-D.bhandari-dhc-class-based-


        prefi=
x] and [I-D.korhonen-6man-prefix-properties], thus reducing


        the amount of c=
ontext maintained in the network.


 

--_000_6E31144C030982429702B11D6746B98C370CBF59szxeml557mbxchi_-- From h.anthony.chan@huawei.com Sat Sep 28 08:28:44 2013 Return-Path: X-Original-To: dmm@ietfa.amsl.com Delivered-To: dmm@ietfa.amsl.com Received: from localhost (localhost [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id 9D14321E8051 for ; Sat, 28 Sep 2013 08:28:44 -0700 (PDT) X-Virus-Scanned: amavisd-new at amsl.com X-Spam-Flag: NO X-Spam-Score: -6.546 X-Spam-Level: X-Spam-Status: No, score=-6.546 tagged_above=-999 required=5 tests=[AWL=0.053, BAYES_00=-2.599, HTML_MESSAGE=0.001, RCVD_IN_DNSWL_MED=-4] Received: from mail.ietf.org ([12.22.58.30]) by localhost (ietfa.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id j-sTRF6EnDcp for ; Sat, 28 Sep 2013 08:28:40 -0700 (PDT) Received: from lhrrgout.huawei.com (lhrrgout.huawei.com [194.213.3.17]) by ietfa.amsl.com (Postfix) with ESMTP id 682D821E8090 for ; Sat, 28 Sep 2013 08:28:37 -0700 (PDT) Received: from 172.18.7.190 (EHLO lhreml203-edg.china.huawei.com) ([172.18.7.190]) by lhrrg01-dlp.huawei.com (MOS 4.3.7-GA FastPath queued) with ESMTP id AYJ39022; Sat, 28 Sep 2013 15:28:36 +0000 (GMT) Received: from LHREML403-HUB.china.huawei.com (10.201.5.217) by lhreml203-edg.huawei.com (172.18.7.221) with Microsoft SMTP Server (TLS) id 14.3.146.0; Sat, 28 Sep 2013 16:28:35 +0100 Received: from SZXEML423-HUB.china.huawei.com (10.82.67.162) by lhreml403-hub.china.huawei.com (10.201.5.217) with Microsoft SMTP Server (TLS) id 14.3.146.0; Sat, 28 Sep 2013 16:28:34 +0100 Received: from szxeml557-mbx.china.huawei.com ([169.254.5.217]) by szxeml423-hub.china.huawei.com ([10.82.67.162]) with mapi id 14.03.0146.000; Sat, 28 Sep 2013 23:28:31 +0800 From: h chan To: "Sri Gundavelli (sgundave)" , "dmm@ietf.org" Thread-Topic: [DMM] I-D Action: draft-ietf-dmm-requirements-07.txt Thread-Index: AQHOogjoa3/L9muNGkS5wozrKtEQEpnXRZkQgABo14CAAJoKEIADLS4wgAAEQpA= Date: Sat, 28 Sep 2013 15:28:29 +0000 Message-ID: <6E31144C030982429702B11D6746B98C370CBF5F@szxeml557-mbx.china.huawei.com> References: <1377486056.79906.YahooMailNeo@web163805.mail.gq1.yahoo.com> <24C0F3E22276D9438D6F366EB89FAEA8116A4D4F@xmb-aln-x03.cisco.com> Accept-Language: en-US, zh-CN Content-Language: en-US X-MS-Has-Attach: X-MS-TNEF-Correlator: x-originating-ip: [10.46.64.17] Content-Type: multipart/alternative; boundary="_000_6E31144C030982429702B11D6746B98C370CBF5Fszxeml557mbxchi_" MIME-Version: 1.0 X-CFilter-Loop: Reflected Subject: Re: [DMM] I-D Action: draft-ietf-dmm-requirements-07.txt X-BeenThere: dmm@ietf.org X-Mailman-Version: 2.1.12 Precedence: list List-Id: Distributed Mobility Management Working Group List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Sat, 28 Sep 2013 15:28:44 -0000 --_000_6E31144C030982429702B11D6746B98C370CBF5Fszxeml557mbxchi_ Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: quoted-printable Sri, It appears that version 09 has attempted to address all your comments. Plea= se check. H Anthony Chan From: h chan Sent: Saturday, September 28, 2013 10:25 AM To: 'Sri Gundavelli (sgundave)'; 'dmm@ietf.org' Subject: RE: [DMM] I-D Action: draft-ietf-dmm-requirements-07.txt Sri, We have revised the Introduction Section and PS1 in version 09. Our respons= es are in-line. The following have contributed to the above revisions in draft 09: Pierrick= , Dapeng, and Jouni From: dmm-bounces@ietf.org [mailto:dmm-bounces= @ietf.org] On Behalf Of Sri Gundavelli (sgundave) Sent: Sunday, August 25, 2013 10:04 PM To: dmm@ietf.org Subject: Re: [DMM] I-D Action: draft-ietf-dmm-requirements-07.txt Please see inline for some comments. Regards Sri ....... 4. Problem Statement The problems that can be addressed with DMM are summarized in the following: PS1: Non-optimal routes Routing via a centralized anchor often results in a longer route. [Sri] "longer route" ? I assume this is about routing/tx delay. Please re-= word. Accept and revised in version 08 The problem is manifested, for example, when accessing a local server or servers of a Content Delivery Network (CDN), or when receiving locally available IP multicast or sending IP multicast packets. [Sri] Does RFC 6705, or RFC 6909 does not address this issue ? May be the C= DN example is incorrect. RFC6909 is clarified further in the Introduction Section in version 09. (1) Mobile users are, more than ever, consuming Internet content; such traffic imposes new requirements on mobile core networks for data traffic delivery. The presence of content providers closer to Internet Service Providers (ISP) network requires taking into account local Content Delivery Networks (CDNs) while providing mobility services. Moreover, when the traffic demand exceeds available capacity, service providers need to implement new strategies such as selective IPv4 traffic offload (e.g. [RFC6909], 3GPP work items LIPA/SIPTO [TS.23.401]) through alternative access networks (e.g. WLAN) [Paper- Mobile.Data.Offloading]. RFC6705 is clarified under PS1 in version 09 PS1: Non-optimal routes Routing via a centralized anchor often results in non-optimal routes, thereby increasing the end-to-end delay. The problem is manifested, for example, when accessing a nearby server or servers of a Content Delivery Network (CDN), or when receiving locally available IP multicast or sending IP multicast packets. (Existing route optimization is only a host-based solution. On the other hand, localized routing with PMIPv6 [RFC6705] addresses only a part of the problem where both the MN and the CN are located in the PMIP domain and attached to a MAG, and is not applicable when the CN is outside the PMIP domain or does not behave like an MN.) PS2: Divergence from other evolutionary trends in network architectures such as distribution of content delivery. Centralized mobility management can become non-optimal with a flat network architecture. [Sri] How is this making the case of DMM ? We want the MN to access content= locally in the access network and we want localized routing. We have that = in the form of 6705 and 6909. The other approach is give localized IP addre= sses and have the content locally accessed. There are simply two many consi= derations and points may be valid, but when we bring all those assumptions.= We are bringing these points in a less logical manner without stating the = assumptions. Our response is similar to that in PS1. Also PS2 is referring to flat mobil= e network. PS3: Low scalability of centralized tunnel management and mobility context maintenance Setting up tunnels through a central anchor and maintaining mobility context for each MN usually requires more concentrated resources in a centralized design, thus reducing scalability. Distributing the tunnel maintenance function and the mobility context maintenance function among different network entities with proper signaling protocol design can increase scalability. PS4: Single point of failure and attack Centralized anchoring designs may be more vulnerable to single points of failures and attacks than a distributed system. The impact of a successful attack on a system with centralized mobility management can be far greater as well. PS5: Unnecessarily reserving resources to provide mobility support to nodes that do not need such support IP mobility support is not always required, and not every parameter of mobility context is always used. For example, some applications do not need a stable IP address during a handover to maintain session continuity. Sometimes, the entire application session runs while the terminal does not change the point of attachment. Besides, some sessions, e.g. SIP-based sessions, can handle mobility at the application layer and hence do not need IP mobility support; it is then more efficient to deactivate IP mobility support for such sessions. [Sri] Mobility systems today do support the aspect of service for the subs= criber, as "Simple IP", or "Mobile IP". A PDSN can assign a local IP addres= s and it does not have to be the home address. Network does have this intel= ligence, but what is missing is the client's ability to pick the correct ty= pe of IP address, among different IP addresses. The network is also the mis= sing the aspect of marking those addresses with proper properties. The curr= ently active drafts in IETF are to address this issue. We should talk about= these missing semantics. draft-bhandari-dhc-class-based-prefix-05 draft-korhonen-6man-prefix-properties-02 These references have been added in the Introduction Section in version 09 (2) Today's mobile networks present service providers with new challenges. Mobility patterns indicate that mobile nodes often remain attached to the same point of attachment for considerable periods of time [Paper-Locating.User]. Specific IP mobility management support is not required for applications that launch and complete their sessions while the mobile node is connected to the same point of attachment. However, currently, IP mobility support is designed for always-on operation, maintaining all parameters of the context for each mobile subscriber for as long as they are connected to the network. This can result in a waste of resources and unnecessary costs for the service provider. Infrequent node mobility coupled with application intelligence suggest that mobility support could be provided selectively such as in [I-D.bhandari-dhc-class-based- prefix] and [I-D.korhonen-6man-prefix-properties], thus reducing the amount of context maintained in the network. --_000_6E31144C030982429702B11D6746B98C370CBF5Fszxeml557mbxchi_ Content-Type: text/html; charset="us-ascii" Content-Transfer-Encoding: quoted-printable

Sri,

It appears that version 0= 9 has attempted to address all your comments. Please check.

 <= /p>

H Anthony Chan=

 <= /p>

From: h chan
Sent: Saturday, September 28, 2013 10:25 AM
To: 'Sri Gundavelli (sgundave)'; 'dmm@ietf.org'
Subject: RE: [DMM] I-D Action: draft-ietf-dmm-requirements-07.txt

 

Sri,

 <= /p>

We have revised the Intro= duction Section and PS1 in version 09. Our responses are in-line.

 <= /p>

The following have contri= buted to the above revisions in draft 09: Pierrick, Dapeng, and Jouni<= /o:p>

 <= /p>

 <= /p>

From: dmm-bounces@ietf.org [mailto:dmm-bounces@ietf.org] On Behalf Of Sri Gundavelli (sgundave)
Sent: Sunday, August 25, 2013 10:04 PM
To: dmm@ietf.org
Subject: Re: [DMM] I-D Action: draft-ietf-dmm-requirements-07.txt

 

Please see inline for some comments.

 

Regards

Sri

 

…….
 
4.  Problem St=
atement
 
   The pr=
oblems that can be addressed with DMM are summarized in the
   follow=
ing:
 
   PS1:&n=
bsp; Non-optimal routes
 
   &=
nbsp;     Routing via a centralized anchor often result=
s in a longer
   &=
nbsp;     route.  
[Sri] "longer route" ?&=
nbsp; I assume this is about routing/tx delay. Please re-word.
 
Accept and revised in=
 version 08
 
The problem is mani=
fested, for example, when accessing
   &=
nbsp;     a local server or servers of a Content Delive=
ry Network (CDN),
   &=
nbsp;     or when receiving locally available IP multic=
ast or sending IP
   &=
nbsp;     multicast packets.
 
[Sri] Does RFC 67=
05, or RFC 6909 does not address this issue ? May be the CDN example is inc=
orrect. 
 
RFC6909 is clarified =
further in the Introduction Section in version 09.  =

   (1)  Mobile users are, more than ever, c= onsuming Internet content;

        such traffic im= poses new requirements on mobile core networks

        for data traffi= c delivery.  The presence of content providers

        closer to Inter= net Service Providers (ISP) network requires

        taking into acc= ount local Content Delivery Networks (CDNs) while

        providing mobil= ity services.  Moreover, when the traffic demand

        exceeds availab= le capacity, service providers need to implement

        new strategies = such as selective IPv4 traffic offload (e.g.

        [RFC6909], 3GPP= work items LIPA/SIPTO [TS.23.401]) through

        alternative acc= ess networks (e.g.  WLAN) [Paper-

        =
Mobile.Data.Offloading].  
 =

RFC6705 is clarified=
 under PS1 in version 09

   PS1:  Non-optimal routes

 

         Routing v= ia a centralized anchor often results in non-optimal

         routes, t= hereby increasing the end-to-end delay.  The problem=

         is manife= sted, for example, when accessing a nearby server or

         servers o= f a Content Delivery Network (CDN), or when receiving

         locally a= vailable IP multicast or sending IP multicast packets.

        = (Existing route optimization is only a host-based solution.  On<= /o:p>

        = the other hand, localized routing with PMIPv6 [RFC6705]<= /p>

        = addresses only a part of the problem where both the MN and the<= /span>

        = CN are located in the PMIP domain and attached to a MAG, and is=

        = not applicable when the CN is outside the PMIP domain or does

        = not behave like an MN.)

 
 
 
   PS2:&n=
bsp; Divergence from other evolutionary trends in network=

   &=
nbsp;     architectures such as distribution of content=
 delivery.
 
   &=
nbsp;     Centralized mobility management can become no=
n-optimal with a
   &=
nbsp;     flat network architecture.<=
/pre>

 


[Sri] How is this=
 making the case of DMM ? We want the MN to access content locally in the a=
ccess network and we want localized routing. We have that in the form of 67=
05 and 6909. The other approach is give localized IP addresses and have the=
 content locally accessed. There are simply two many considerations and poi=
nts may be valid, but when we bring all those assumptions. We are bringing =
these points in a less logical manner without stating the assumptions.


 =



Our response is simi=
lar to that in PS1. Also PS2 is referring to flat mobile network. 


 


   PS3:&n=
bsp; Low scalability of centralized tunnel management and mobility


   &=
nbsp;     context maintenance


 


   &=
nbsp;     Setting up tunnels through a central anchor a=
nd maintaining


   &=
nbsp;     mobility context for each MN usually requires=
 more concentrated


   &=
nbsp;     resources in a centralized design, thus reduc=
ing scalability.


   &=
nbsp;     Distributing the tunnel maintenance function =
and the mobility


   &=
nbsp;     context maintenance function among different =
network entities


   &=
nbsp;     with proper signaling protocol design can inc=
rease scalability.


 


   PS4:&n=
bsp; Single point of failure and attack


 


   &=
nbsp;     Centralized anchoring designs may be more vul=
nerable to single


   &=
nbsp;     points of failures and attacks than a distrib=
uted system.  The


   &=
nbsp;     impact of a successful attack on a system wit=
h centralized


   &=
nbsp;     mobility management can be far greater as wel=
l.


 


   PS5:&n=
bsp; Unnecessarily reserving resources to provide mobility support


   &=
nbsp;     to nodes that do not need such support


 


   &=
nbsp;     IP mobility support is not always required, a=
nd not every


   &=
nbsp;     parameter of mobility context is always used.=
  For example,


   &=
nbsp;     some applications do not need a stable IP add=
ress during a


   &=
nbsp;     handover to maintain session continuity. =
; Sometimes, the entire


   &=
nbsp;     application session runs while the terminal d=
oes not change the


   &=
nbsp;     point of attachment.  Besides, some sess=
ions, e.g.  SIP-based


   &=
nbsp;     sessions, can handle mobility at the applicat=
ion layer and


   &=
nbsp;     hence do not need IP mobility support; it is =
then more


   &=
nbsp;     efficient to deactivate IP mobility support f=
or such sessions.


 


[Sri]  Mobility systems toda=
y do support the aspect of service for the subscriber, as "Simple IP&q=
uot;, or "Mobile IP". A PDSN can assign a local IP address and it=
 does not have to be the home address. Network does have this intelligence,=
 but what is missing is the client's ability to pick the correct type of IP=
 address, among different IP addresses. The network is also the missing the=
 aspect of marking those addresses with proper properties. The currently ac=
tive drafts in IETF are to address this issue. We should talk about these m=
issing semantics.<=
/pre>

draft-bhandari-dhc-class-based-prefix-05  


draft-korhonen-6man-prefix-properties-02<=
/span>


 


These references have=
 been added in the Introduction Section in version 09

 


   (2)  Today's mobile networks present ser=
vice providers with new


        challenges.&nbs=
p; Mobility patterns indicate that mobile nodes often


        remain attached=
 to the same point of attachment for considerable


        periods of time=
 [Paper-Locating.User].  Specific IP mobility


        management supp=
ort is not required for applications that launch


        and complete th=
eir sessions while the mobile node is connected


        to the same poi=
nt of attachment.  However, currently, IP


        mobility suppor=
t is designed for always-on operation,


        maintaining all=
 parameters of the context for each mobile


        subscriber for =
as long as they are connected to the network.


        This can result=
 in a waste of resources and unnecessary costs


        for the service=
 provider.  Infrequent node mobility coupled with

        application int=
elligence suggest that mobility support could be


        provided select=
ively
such as in [I-D.bhandari-dhc-class-based-


        prefi=
x] and [I-D.korhonen-6man-prefix-properties], thus reducing


        the amount of c=
ontext maintained in the network.


 

 

--_000_6E31144C030982429702B11D6746B98C370CBF5Fszxeml557mbxchi_--