
From david.a.mcgrew@mindspring.com  Mon Aug  1 10:25:27 2011
Return-Path: <david.a.mcgrew@mindspring.com>
X-Original-To: tsv-area@ietfa.amsl.com
Delivered-To: tsv-area@ietfa.amsl.com
Received: from localhost (localhost [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id E8AFE21F8B86; Mon,  1 Aug 2011 10:25:27 -0700 (PDT)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: 1.989
X-Spam-Level: *
X-Spam-Status: No, score=1.989 tagged_above=-999 required=5 tests=[BAYES_00=-2.599, FRT_STOCK2=3.988, J_CHICKENPOX_64=0.6]
Received: from mail.ietf.org ([64.170.98.30]) by localhost (ietfa.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id fvUBOL-Psra4; Mon,  1 Aug 2011 10:25:26 -0700 (PDT)
Received: from elasmtp-galgo.atl.sa.earthlink.net (elasmtp-galgo.atl.sa.earthlink.net [209.86.89.61]) by ietfa.amsl.com (Postfix) with ESMTP id B1B0221F8B44; Mon,  1 Aug 2011 10:25:26 -0700 (PDT)
Received: from [69.251.37.177] (helo=stealth-10-32-254-211.cisco.com) by elasmtp-galgo.atl.sa.earthlink.net with esmtpsa (TLSv1:AES128-SHA:128) (Exim 4.67) (envelope-from <david.a.mcgrew@mindspring.com>) id 1QnwEs-0005jI-OP; Mon, 01 Aug 2011 13:25:14 -0400
Message-Id: <4884218B-9477-4F6D-84BB-9EB07F4D5CA9@mindspring.com>
From: "David A. McGrew" <david.a.mcgrew@mindspring.com>
To: Andrea Bittau <bittau@cs.stanford.edu>
In-Reply-To: <20110730014544.GM3751@shorty>
Content-Type: text/plain; charset=US-ASCII; format=flowed; delsp=yes
Content-Transfer-Encoding: 7bit
Mime-Version: 1.0 (Apple Message framework v936)
Subject: Re: [saag] fyi: tcpcrypt draft
Date: Mon, 1 Aug 2011 10:25:14 -0700
References: <4DC6C56D.6040709@ieca.com> <D1CCED3D-C041-4DA2-B999-D3EF7FC889D0@cisco.com> <20110727184256.GA3690@shorty> <8A00B447-53C3-44D0-8238-8DAAE44688CB@mindspring.com> <20110730014544.GM3751@shorty>
X-Mailer: Apple Mail (2.936)
X-ELNK-Trace: ad1f9a46c4c7bfd19649176a89d694c0f43c108795ac4507c408d7d3d2e9aa2d4f5548039267e14d350badd9bab72f9c350badd9bab72f9c350badd9bab72f9c
X-Originating-IP: 69.251.37.177
X-Mailman-Approved-At: Mon, 01 Aug 2011 11:34:56 -0700
Cc: tcpcrypt@scs.stanford.edu, tsv-area@ietf.org, saag@ietf.org
X-BeenThere: tsv-area@ietf.org
X-Mailman-Version: 2.1.12
Precedence: list
List-Id: IETF Transport and Services Area Mailing List <tsv-area.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/options/tsv-area>, <mailto:tsv-area-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/tsv-area>
List-Post: <mailto:tsv-area@ietf.org>
List-Help: <mailto:tsv-area-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/tsv-area>, <mailto:tsv-area-request@ietf.org?subject=subscribe>
X-List-Received-Date: Mon, 01 Aug 2011 17:25:28 -0000

Hi Andrea,

thanks for the detail.

On Jul 29, 2011, at 6:45 PM, Andrea Bittau wrote:

> On Thu, Jul 28, 2011 at 07:33:05AM -0700, David A. McGrew wrote:
>> A lot of work has been done on using IPsec to solve these problems,
>> and it is important to explain in detail how tcpcrypt improves on
>> this work.  Besides the opportunistic encryption effort, there is
>> BTNS (draft-ietf-btns-channel-binding-api-00 defines channel
>> bindings for IPsec, for instance), RFC 3947 for running IPsec over
>> NAT.  (Half joking: would the configuration of
>> draft-ietf-ipsec-internet-key-00 with NAT-T solve all or most these
>> problems?)
>
> tcpcrypt makes incremental deployment easier because probing is built
> into existing packets and fallback comes at no extrea cost.   When a  
> SYN
> packet is sent out and tcpcrypt is not supported at the other end, we
> can fallback to regular tcpcrypt at no extra cost - just carry on  
> with a
> vanilla TCP handshake.  How would this work with BTNS or IPSec?  Do  
> you
> probe for IPSec (and then probe for NAT-T), wait for a timeout, and  
> then
> send a TCP SYN?  If so, it adds network overhead (extra packets) and
> latency (probe timeout).

Yes, valid points.  It would be good to add this motivating text to  
the draft.

>
> tcpcrypt also has less protocol (space) overhead.  NAT-T requires
> encapsulating IP in UDP, and NAT seems like a common deployment
> scenario.  It also requires opening more ports on a firewall and
> configuring additional services (NAT-T).
>
>> What TCP applications can't use TLS?   Do you mean that some apps
>> don't use TLS in current implementations, or that there is some
>> basic incompatibility?
>
> There's a bootstrap problem.  If I have SMTP on TCP/25 I have to  
> assume
> it's clear-text (I can't send an SSL handshake packet).  Thankfully,
> SMTP is an extensible protocol and I can send STARTTLS and then go on
> with SSL.  If the protocol didn't permit any "options" / commands, I'd
> be stuck to plain-text.  I'd have to allocate a new port.  That makes
> incremental deployment harder - do I send two SYNs?

Got it.  I would state it this way: while TLS can be used to protect  
any TCP application in principle, in practice it is not always  
available, and introducing it into an application protocol requires a  
mechanism for advertising and selecting to use TLS where it is  
available, and falling back on unprotected communications where it is  
not available.

>
>> What about running TCP on top of DTLS?   That would protect the TCP
>> session, while keeping the crypto separate from the TCP stack.
>
> Same answer as per BTNS and NAT-T - we can efficiently incrementally
> deploy and have less protocol overhead.
>
>> 1.  How easy to configure will tcpcrypt be once it supports mutual
>> authentication, the enforcement of configurable security policies,
>> and crypto algorithm agility (e.g. ECDH and ECDSA)?
>
> Authentication is left to the application, so tcpcrypt will support
> authentication only if the application supports it.  If the  
> application
> supports it, then a mechanism is already in place, e.g., htpasswd in
> Apache for passwords.  So in the case of Apache, it'll include
> tcpcrypt's session id when doing http digest authentication, for
> example.
>
> We're also working on an authentication library that attempts to
> authenticate tcpcrypt using DNSSEC+DANE that can be used by all apps  
> if
> DANE is deployed.  The configuration will require storing a  
> certificate
> fingerprint in DNS.
>
> We expect that most applications (e.g., the ones running clear-text
> today) won't care about the crypto algorithms used in tcpcrypt's
> handshake.  So the security policy will be set system-wide (e.g.,  
> allow
> only RSA, SHA1 or whatever).  For security concious applications it'll
> be a setsockopt and presumably those applications already have a
> configuration file for whatever the already use today for crypto.
>
>> 2.  The comparison of tcpcrypt to TLS is only considering the RSA
>> ciphersuites in the latter protocol.  The efficiency and security
>> properties of other ciphersuites are quite different.
>
> Yes, but we expect tcpcrypt to be no worse (or better) than TLS.

Agreed; but note that this is a far different statement than "36 x  
better performance", which if I recall right was stated in the  
presentation at IETF81.  It is important to provide a detailed  
comparison, especially to the non-crypto audience who might miss the  
full implications.


> tcpcrypt fundamentally does however have lower latency for conncection
> establishment compared to TLS due to fewer RTTs needed in  
> negotiating a
> key (we leverage the 3-way handshake and add only an extra leg),
> especially when session caching.
>
>> 3.  Reversing the client and server public/private key operations in
>> tcpcrypt, which is used to make the protocol more scalable, is
>> applicable only to the RSA algorithm.   In particular, the situation
>> with elliptic curve cryptography is quite different.  This raises
>> some questions: how does tcpcrypt with RSA compare to an ECDH/ECDSA
>> TLS ciphersuite, say, at security levels of 128, 192, or 256 bits?
>> How does tcpcrypt perform with other algorithms?   Given the
>> interest in ECC as a future direction for cryptography, these
>> questions deserve some thought.
>
> Yes, we'll implement those but we do not expect tcpcrypt's performance
> to be much different from a TLS implementation - we both do the same
> thing (e.g., ECDH).

What will be especially interesting is to see how the reversal of RSA  
private/public key operations compares to ECC as a function of  
security level.

>
>> What you are describing here is true for the RSA ciphersuites in
>> SSL, but is not true for the Diffie-Hellman ciphersuites, or the
>> ECDH ones (see Section 7.4.2 of RFC 5246 for instance).
>
> tcpcrypyt's main goal is to be deployed ubiquitously.  Improved
> performance in one case is only a bonus.  It has a bunch of other
> advantages (single integrated mechanism, incrementally deployable)  
> apart
> from RSA performance.
>
>> Why is graceful fallback easier at the TCP layer than at another
>> layer?   And if there is some mechanism within TCP that facilitates
>> fallback, would it be possible to use that mechanism in conjunction
>> with a cryptographic protocol external to TCP?  That last option
>> would minimize the impact to TCP implementations, meaning less new
>> code inside kernels, and fewer states, and so on.
>
> Gracefully, and efficient, I should add.  The problem is detecting
> whether the other end supports crypto mechanism X.  At a below- 
> transport
> layer, we'd have to send our IP packet, wait for a timeout, and then
> send a SYN.  At a transport layer we're sending out a SYN so we can
> easily piggyback options (perfect!).  At an application layer what's  
> the
> first thing we send?  STARTTLS?  Does the application-layer protocol
> allow such new commands or will it drop the command?
>
> The minimum TCP mechanism needed for a graceful and efficient fallback
> is a TCP option in the SYN.  So, tcpcrypt-lite would modify a kernel
> with a setsockopt(CRYPTO_OPTION) to include the CRYPTO option in a SYN
> and getsockopt(CRYPTO_OPTION) to see if the CRYPTO option was echoed  
> in
> the SYN-ACK.  At that point SSL_accept() can be called.

This sounds like an option worth pursuing, no pun intended.  You make  
good points about why TCP is a good place to do the capability  
discovery and fallback for crypto.   Why not define a TCP option that  
can enable devices to discover TLS or tcpcrypt or other higher-layer  
crypto protocols?   That option would be much easier to implement and  
standardize, and it could tie into a full tcpcrypt protocol that  
resides in a shim above the TCP layer.  The separation will create  
useful implementation alternatives, and will isolate the TCP state  
machine from changes.

For concreteness, I'm thinking of a TCP option that looks like this:

                +---------+---------+-------------------+
                |  Kind   |Length=2 |     SECCAP        |
                +---------+---------+-------------------+

The Security Capabilities (SECCAP) field contains an unsigned integer  
that indicates
the cryptographic security protocol, if any, the endpoint can use to  
protect subsequent
communications.

   SECCAP Value     Meaning

         0          Reserved for future use
         1          No security protocol is available
         2          TCPcrypt
         3          TLS
         ...        ...



>  But we lose all
> TCP header MACing - e.g., someone can RST our connection.  TCP AO
> suggests that MACing TCP headers is important so crypto protocols  
> should
> do that (hence why we live in the transport layer).

True, but why not design tcpcrypt so that it make use of TCP-AO?    
That seems like an approach that would work with the separation  
outlined above.

> Also, we can
> leverage the existing 3-way handshake to get a head start on crypto
> negotiations.  Sesison caching completes within the three way  
> handshake.
>
>> If the RSA keypair is generated using a pseudorandom process (and
>> there are standards for doing that), the min entropy could be lower.
>>
>> But if it is adequate to have the entropy show up in the nonces, the
>> min-entropy of the key would be a moot point.
>
> The pseudo-random generator has to have enough state to generate prime
> numbers with min entropy 256.  Depending on the prime number  
> generator,
> this might, for instance, require something on the order of 256 + ceil
> (log_2 (log_e (2 ^ 1024))) or 266 bits.  We assume most RSA
> implementations would have more random state than that.

Unfortunately that is not always a safe assumption.  There are  
standards for RSA key generation (see for instance NIST SP 800 56-B  
Section 6.3.1) in which n bits of random data are used to seed a  
pseudorandom process that selects the primes used as the private key;  
for 1024-bit RSA, n will be about 80.

I am confident that tcpcrypt can be made to work around this fact, but  
it may be an important detail (in the security proof if not in  
practice).

David

> _______________________________________________
> saag mailing list
> saag@ietf.org
> https://www.ietf.org/mailman/listinfo/saag


From bittau@gmail.com  Mon Aug  1 12:15:43 2011
Return-Path: <bittau@gmail.com>
X-Original-To: tsv-area@ietfa.amsl.com
Delivered-To: tsv-area@ietfa.amsl.com
Received: from localhost (localhost [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id 5092421F8DE9; Mon,  1 Aug 2011 12:15:43 -0700 (PDT)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -2.991
X-Spam-Level: 
X-Spam-Status: No, score=-2.991 tagged_above=-999 required=5 tests=[AWL=0.608,  BAYES_00=-2.599, RCVD_IN_DNSWL_LOW=-1]
Received: from mail.ietf.org ([64.170.98.30]) by localhost (ietfa.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id oRaVftl0ncCV; Mon,  1 Aug 2011 12:15:42 -0700 (PDT)
Received: from mail-pz0-f53.google.com (mail-pz0-f53.google.com [209.85.210.53]) by ietfa.amsl.com (Postfix) with ESMTP id B5DDF21F8D9A; Mon,  1 Aug 2011 12:15:42 -0700 (PDT)
Received: by pzk6 with SMTP id 6so11133451pzk.26 for <multiple recipients>; Mon, 01 Aug 2011 12:15:49 -0700 (PDT)
DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=gmail.com; s=gamma; h=sender:date:from:to:cc:subject:message-id:references:mime-version :content-type:content-disposition:in-reply-to:x-echelon:user-agent; bh=z2MD7wkTQHlx/P/Je6Nn7x7KoBfi0S6f894FqsZd9rs=; b=qxMwR/FrdNFcpnb3alIe/OVGomR8HiOwAmx+gISy/dl3cGfvy/UFhGGRJxo5ejjZlc WjNCUFqaPwUVxw9y47qgWJYguxB4M/cLgXOL4GE4JTf3NYrQmLWjjyWBnZGgCWbrdIbf ZOCa4GuaBSYsyeM2yNQBrMt+N9P88yp5ES9do=
Received: by 10.68.1.135 with SMTP id 7mr8720500pbm.53.1312226149663; Mon, 01 Aug 2011 12:15:49 -0700 (PDT)
Received: from tapir.cs.ucl.ac.uk (sorbo.scs.stanford.edu [171.66.3.100]) by mx.google.com with ESMTPS id d3sm3178253pbg.12.2011.08.01.12.15.47 (version=TLSv1/SSLv3 cipher=OTHER); Mon, 01 Aug 2011 12:15:48 -0700 (PDT)
Sender: Andrea Bittau <bittau@gmail.com>
Received: by tapir.cs.ucl.ac.uk (Postfix, from userid 0) id 808584810171; Mon,  1 Aug 2011 12:14:07 -0700 (PDT)
Date: Mon, 1 Aug 2011 12:14:07 -0700
From: Andrea Bittau <bittau@cs.stanford.edu>
To: "David A. McGrew" <david.a.mcgrew@mindspring.com>
Subject: Re: [saag] fyi: tcpcrypt draft
Message-ID: <20110801191407.GA3765@shorty>
References: <4DC6C56D.6040709@ieca.com> <D1CCED3D-C041-4DA2-B999-D3EF7FC889D0@cisco.com> <20110727184256.GA3690@shorty> <8A00B447-53C3-44D0-8238-8DAAE44688CB@mindspring.com> <20110730014544.GM3751@shorty> <4884218B-9477-4F6D-84BB-9EB07F4D5CA9@mindspring.com>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <4884218B-9477-4F6D-84BB-9EB07F4D5CA9@mindspring.com>
X-Echelon: Bush Bomb War KGB
User-Agent: Mutt/1.5.21 (2010-09-15)
Cc: tcpcrypt@scs.stanford.edu, tsv-area@ietf.org, saag@ietf.org
X-BeenThere: tsv-area@ietf.org
X-Mailman-Version: 2.1.12
Precedence: list
List-Id: IETF Transport and Services Area Mailing List <tsv-area.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/options/tsv-area>, <mailto:tsv-area-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/tsv-area>
List-Post: <mailto:tsv-area@ietf.org>
List-Help: <mailto:tsv-area-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/tsv-area>, <mailto:tsv-area-request@ietf.org?subject=subscribe>
X-List-Received-Date: Mon, 01 Aug 2011 19:15:43 -0000

On Mon, Aug 01, 2011 at 10:25:14AM -0700, David A. McGrew wrote:
> This sounds like an option worth pursuing, no pun intended.  You
> make good points about why TCP is a good place to do the capability
> discovery and fallback for crypto.   Why not define a TCP option
> that can enable devices to discover TLS or tcpcrypt or other
> higher-layer crypto protocols?   That option would be much easier to
> implement and standardize, and it could tie into a full tcpcrypt
> protocol that resides in a shim above the TCP layer.  The separation
> will create useful implementation alternatives, and will isolate the
> TCP state machine from changes.

We have been considering this option throughout the work (and will still
do) but:

* No TCP header protection.  TCP-AO, which you suggest, does not
  interoperate with NATs.  We want something that works "out of the
  box".

* We can't leverage the 3-way handshake if operating at the application
  layer.  With tcpcrypt, we can do all session key negotiation in a
  4-way handshake (just one extra leg), or 3 if session caching.  Saving
  RTTs seems like a good thing.

>From a clean-slate perspective, it seems that integrating this into TCP
has all the advantages we're looking for.  Doing minimal work at TCP
(like a signle "do you crypto?" option) and doing the rest at an
application layer seems to come at a compromise (e.g., the drawbacks
listed above).

>From an implementation perspective, I agree that the least changes to
TCP are probably the best option.  tcpcrypt, at least for us, seems to
have a good balance between clean-slate and (relatively) few changes to
TCP, hence why we're pushing for it, as is.  If there were
non-implementation arguments as to app vs. transport layer we'd love to
hear those.

From touch@isi.edu  Mon Aug  1 13:20:02 2011
Return-Path: <touch@isi.edu>
X-Original-To: tsv-area@ietfa.amsl.com
Delivered-To: tsv-area@ietfa.amsl.com
Received: from localhost (localhost [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id 5B89E21F8D3F; Mon,  1 Aug 2011 13:20:02 -0700 (PDT)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -105.345
X-Spam-Level: 
X-Spam-Status: No, score=-105.345 tagged_above=-999 required=5 tests=[AWL=1.254, BAYES_00=-2.599, RCVD_IN_DNSWL_MED=-4, USER_IN_WHITELIST=-100]
Received: from mail.ietf.org ([64.170.98.30]) by localhost (ietfa.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id XGzTv9-rlQcy; Mon,  1 Aug 2011 13:20:02 -0700 (PDT)
Received: from boreas.isi.edu (boreas.isi.edu [128.9.160.161]) by ietfa.amsl.com (Postfix) with ESMTP id D9E1721F8D18; Mon,  1 Aug 2011 13:20:01 -0700 (PDT)
Received: from [128.9.160.166] (abc.isi.edu [128.9.160.166]) (authenticated bits=0) by boreas.isi.edu (8.13.8/8.13.8) with ESMTP id p71KJgoF027995 (version=TLSv1/SSLv3 cipher=DHE-RSA-AES256-SHA bits=256 verify=NOT); Mon, 1 Aug 2011 13:19:43 -0700 (PDT)
Message-ID: <4E370A5E.1050608@isi.edu>
Date: Mon, 01 Aug 2011 13:19:42 -0700
From: Joe Touch <touch@isi.edu>
User-Agent: Mozilla/5.0 (Windows NT 6.1; WOW64; rv:5.0) Gecko/20110624 Thunderbird/5.0
MIME-Version: 1.0
To: "David A. McGrew" <david.a.mcgrew@mindspring.com>
Subject: Re: [saag] fyi: tcpcrypt draft
References: <4DC6C56D.6040709@ieca.com> <D1CCED3D-C041-4DA2-B999-D3EF7FC889D0@cisco.com> <20110727184256.GA3690@shorty> <8A00B447-53C3-44D0-8238-8DAAE44688CB@mindspring.com> <20110730014544.GM3751@shorty> <4884218B-9477-4F6D-84BB-9EB07F4D5CA9@mindspring.com>
In-Reply-To: <4884218B-9477-4F6D-84BB-9EB07F4D5CA9@mindspring.com>
Content-Type: text/plain; charset=ISO-8859-1; format=flowed
Content-Transfer-Encoding: 7bit
X-ISI-4-43-8-MailScanner: Found to be clean
X-MailScanner-From: touch@isi.edu
Cc: tcpcrypt@scs.stanford.edu, Andrea Bittau <bittau@cs.stanford.edu>, tsv-area@ietf.org, saag@ietf.org
X-BeenThere: tsv-area@ietf.org
X-Mailman-Version: 2.1.12
Precedence: list
List-Id: IETF Transport and Services Area Mailing List <tsv-area.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/options/tsv-area>, <mailto:tsv-area-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/tsv-area>
List-Post: <mailto:tsv-area@ietf.org>
List-Help: <mailto:tsv-area-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/tsv-area>, <mailto:tsv-area-request@ietf.org?subject=subscribe>
X-List-Received-Date: Mon, 01 Aug 2011 20:20:02 -0000

On 8/1/2011 10:25 AM, David A. McGrew wrote:
...
>> But we lose all
>> TCP header MACing - e.g., someone can RST our connection. TCP AO
>> suggests that MACing TCP headers is important so crypto protocols should
>> do that (hence why we live in the transport layer).
>
> True, but why not design tcpcrypt so that it make use of TCP-AO? That
> seems like an approach that would work with the separation outlined above.

TCP-AO could easily support a mode where the TCP body was encrypted, 
e.g., where the encryption algorithm were defined to use the connection 
context (TCP header, IP pseudoheader) as context.

I had been planning on working that into the experimental TCP-AO NAT 
doc, since it could be as useful as the NAT variant.

That won't help initialize the keys, as would be needed before the SYN 
arrives for TCP-AO. That sort of in-band keying was excluded from TCP-AO 
to keep the SYN header size down.

Joe


From touch@isi.edu  Mon Aug  1 17:50:53 2011
Return-Path: <touch@isi.edu>
X-Original-To: tsv-area@ietfa.amsl.com
Delivered-To: tsv-area@ietfa.amsl.com
Received: from localhost (localhost [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id 45CBC1F0C51; Mon,  1 Aug 2011 17:50:53 -0700 (PDT)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -103.323
X-Spam-Level: 
X-Spam-Status: No, score=-103.323 tagged_above=-999 required=5 tests=[AWL=-0.724, BAYES_00=-2.599, USER_IN_WHITELIST=-100]
Received: from mail.ietf.org ([64.170.98.30]) by localhost (ietfa.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id p6e7p9t-dVr4; Mon,  1 Aug 2011 17:50:51 -0700 (PDT)
Received: from darkstar.isi.edu (darkstar.isi.edu [128.9.128.127]) by ietfa.amsl.com (Postfix) with ESMTP id CAC8B1F0C49; Mon,  1 Aug 2011 17:50:48 -0700 (PDT)
Received: from [128.9.160.166] (abc.isi.edu [128.9.160.166]) (authenticated bits=0) by darkstar.isi.edu (8.13.8/8.13.8) with ESMTP id p720oPvM016591 (version=TLSv1/SSLv3 cipher=DHE-RSA-AES256-SHA bits=256 verify=NOT); Mon, 1 Aug 2011 17:50:26 -0700 (PDT)
Message-ID: <4E3749D1.4000403@isi.edu>
Date: Mon, 01 Aug 2011 17:50:25 -0700
From: Joe Touch <touch@isi.edu>
User-Agent: Mozilla/5.0 (Windows NT 6.1; WOW64; rv:5.0) Gecko/20110624 Thunderbird/5.0
MIME-Version: 1.0
To: "David A. McGrew" <david.a.mcgrew@mindspring.com>
Subject: Re: [saag] fyi: tcpcrypt draft
References: <4DC6C56D.6040709@ieca.com> <D1CCED3D-C041-4DA2-B999-D3EF7FC889D0@cisco.com> <20110727184256.GA3690@shorty> <8A00B447-53C3-44D0-8238-8DAAE44688CB@mindspring.com> <20110730014544.GM3751@shorty> <4884218B-9477-4F6D-84BB-9EB07F4D5CA9@mindspring.com> <4E370A5E.1050608@isi.edu> <45875A47-E146-40F9-9B90-96AD015BBC85@mindspring.com>
In-Reply-To: <45875A47-E146-40F9-9B90-96AD015BBC85@mindspring.com>
Content-Type: text/plain; charset=ISO-8859-1; format=flowed
Content-Transfer-Encoding: 7bit
X-ISI-4-43-8-MailScanner: Found to be clean
X-MailScanner-From: touch@isi.edu
Cc: tcpcrypt@scs.stanford.edu, tsv-area@ietf.org, saag@ietf.org
X-BeenThere: tsv-area@ietf.org
X-Mailman-Version: 2.1.12
Precedence: list
List-Id: IETF Transport and Services Area Mailing List <tsv-area.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/options/tsv-area>, <mailto:tsv-area-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/tsv-area>
List-Post: <mailto:tsv-area@ietf.org>
List-Help: <mailto:tsv-area-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/tsv-area>, <mailto:tsv-area-request@ietf.org?subject=subscribe>
X-List-Received-Date: Tue, 02 Aug 2011 00:50:53 -0000

On 8/1/2011 5:37 PM, David A. McGrew wrote:
>
> On Aug 1, 2011, at 1:19 PM, Joe Touch wrote:
>
>>
>>
>> On 8/1/2011 10:25 AM, David A. McGrew wrote:
>> ...
>>>> But we lose all
>>>> TCP header MACing - e.g., someone can RST our connection. TCP AO
>>>> suggests that MACing TCP headers is important so crypto protocols
>>>> should
>>>> do that (hence why we live in the transport layer).
>>>
>>> True, but why not design tcpcrypt so that it make use of TCP-AO? That
>>> seems like an approach that would work with the separation outlined
>>> above.
>>
>> TCP-AO could easily support a mode where the TCP body was encrypted,
>> e.g., where the encryption algorithm were defined to use the
>> connection context (TCP header, IP pseudoheader) as context.
>
> It occurs to me that one of the downsides of having authentication via
> TCP-AO and encryption via TCPcrypt would be that authenticated
> encryption (AEAD) could not be used.

You would not want to cross the streams and use both options AFAICT.

>> I had been planning on working that into the experimental TCP-AO NAT
>> doc, since it could be as useful as the NAT variant.
>>
>> That won't help initialize the keys, as would be needed before the SYN
>> arrives for TCP-AO. That sort of in-band keying was excluded from
>> TCP-AO to keep the SYN header size down.
>>
> Can TCP-AO possibly be used in a situation where the key is established
> after the initial message exchange, but there is no key available to
> both endpoints in the first exchange?

No; the assumption in TCP-AO is that a key exists during the initial 
exchange and AO is required. Keys can be changed later.

It would be trivial to set this up as a null key to start with if your 
goal is as you describe, though.

 > If so, is there a strong security
> requirement that militates against doing that?

Generally yes; it prevents AO from protecting you from SYN attacks.

You'd also need to change AO so that (depending on a flag set at both 
endpoints) you change the key after the SYN or not (e.g., increment the 
keyID or somesuch).

> It would be undesirable
> to have an endpoint set up a session (without authentication) then
> immediately have to tear it down (because when authentication was turned
> on, it became apparent that an active attacker had created or modified
> the messages). But perhaps if the "damage" is well contained and quickly
> discovered, it would be OK?

I don't know if this would be considered OK. I'd say it'd be easier to 
separate it into two steps, as hinted above:

1) an updated AO spec that allows a flag (on a master key, e.g.)
	that requires the keyID to be incremented after the SYN
	and that the first key can be null optionally

2) allow null keys (to enable the first part)

Null keys would expose the endpoint during the period they're used, on 
ports they're used. If the period were small (the SYN), then you could 
limit the damage AFAICT, but I'm not sure whether this would be 
considered sufficient or useful.

Joe


From david.a.mcgrew@mindspring.com  Mon Aug  1 17:37:43 2011
Return-Path: <david.a.mcgrew@mindspring.com>
X-Original-To: tsv-area@ietfa.amsl.com
Delivered-To: tsv-area@ietfa.amsl.com
Received: from localhost (localhost [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id 5FC7A21F8C0A; Mon,  1 Aug 2011 17:37:41 -0700 (PDT)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -0.305
X-Spam-Level: 
X-Spam-Status: No, score=-0.305 tagged_above=-999 required=5 tests=[AWL=2.294,  BAYES_00=-2.599]
Received: from mail.ietf.org ([64.170.98.30]) by localhost (ietfa.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id C2VJlDjHyt-X; Mon,  1 Aug 2011 17:37:39 -0700 (PDT)
Received: from elasmtp-masked.atl.sa.earthlink.net (elasmtp-masked.atl.sa.earthlink.net [209.86.89.68]) by ietfa.amsl.com (Postfix) with ESMTP id 0B02F1F0C3D; Mon,  1 Aug 2011 17:37:36 -0700 (PDT)
Received: from [69.251.37.177] (helo=stealth-10-32-254-211.cisco.com) by elasmtp-masked.atl.sa.earthlink.net with esmtpsa (TLSv1:AES128-SHA:128) (Exim 4.67) (envelope-from <david.a.mcgrew@mindspring.com>) id 1Qo2zP-0001XQ-Fh; Mon, 01 Aug 2011 20:37:43 -0400
Message-Id: <45875A47-E146-40F9-9B90-96AD015BBC85@mindspring.com>
From: "David A. McGrew" <david.a.mcgrew@mindspring.com>
To: Joe Touch <touch@isi.edu>
In-Reply-To: <4E370A5E.1050608@isi.edu>
Content-Type: text/plain; charset=US-ASCII; format=flowed; delsp=yes
Content-Transfer-Encoding: 7bit
Mime-Version: 1.0 (Apple Message framework v936)
Subject: Re: [saag] fyi: tcpcrypt draft
Date: Mon, 1 Aug 2011 17:37:42 -0700
References: <4DC6C56D.6040709@ieca.com> <D1CCED3D-C041-4DA2-B999-D3EF7FC889D0@cisco.com> <20110727184256.GA3690@shorty> <8A00B447-53C3-44D0-8238-8DAAE44688CB@mindspring.com> <20110730014544.GM3751@shorty> <4884218B-9477-4F6D-84BB-9EB07F4D5CA9@mindspring.com> <4E370A5E.1050608@isi.edu>
X-Mailer: Apple Mail (2.936)
X-ELNK-Trace: ad1f9a46c4c7bfd19649176a89d694c0f43c108795ac4507044e77353aa95c2453044fb2346e3d6d350badd9bab72f9c350badd9bab72f9c350badd9bab72f9c
X-Originating-IP: 69.251.37.177
X-Mailman-Approved-At: Mon, 01 Aug 2011 18:48:00 -0700
Cc: tcpcrypt@scs.stanford.edu, tsv-area@ietf.org, saag@ietf.org
X-BeenThere: tsv-area@ietf.org
X-Mailman-Version: 2.1.12
Precedence: list
List-Id: IETF Transport and Services Area Mailing List <tsv-area.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/options/tsv-area>, <mailto:tsv-area-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/tsv-area>
List-Post: <mailto:tsv-area@ietf.org>
List-Help: <mailto:tsv-area-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/tsv-area>, <mailto:tsv-area-request@ietf.org?subject=subscribe>
X-List-Received-Date: Tue, 02 Aug 2011 00:37:44 -0000

On Aug 1, 2011, at 1:19 PM, Joe Touch wrote:

>
>
> On 8/1/2011 10:25 AM, David A. McGrew wrote:
> ...
>>> But we lose all
>>> TCP header MACing - e.g., someone can RST our connection. TCP AO
>>> suggests that MACing TCP headers is important so crypto protocols  
>>> should
>>> do that (hence why we live in the transport layer).
>>
>> True, but why not design tcpcrypt so that it make use of TCP-AO? That
>> seems like an approach that would work with the separation outlined  
>> above.
>
> TCP-AO could easily support a mode where the TCP body was encrypted,  
> e.g., where the encryption algorithm were defined to use the  
> connection context (TCP header, IP pseudoheader) as context.

It occurs to me that one of the downsides of having authentication via  
TCP-AO and encryption via TCPcrypt would be that authenticated  
encryption (AEAD) could not be used.

>
> I had been planning on working that into the experimental TCP-AO NAT  
> doc, since it could be as useful as the NAT variant.
>
> That won't help initialize the keys, as would be needed before the  
> SYN arrives for TCP-AO. That sort of in-band keying was excluded  
> from TCP-AO to keep the SYN header size down.
>

Can TCP-AO possibly be used in a situation where the key is  
established after the initial message exchange, but there is no key  
available to both endpoints in the first exchange?   If so, is there a  
strong security requirement that militates against doing that?   It  
would be undesirable to have an endpoint set up a session (without  
authentication) then immediately have to tear it down (because when  
authentication was turned on, it became apparent that an active  
attacker had created or modified the messages).   But perhaps if the  
"damage" is well contained and quickly discovered, it would be OK?

David

> Joe
>
> _______________________________________________
> saag mailing list
> saag@ietf.org
> https://www.ietf.org/mailman/listinfo/saag


From david.black@emc.com  Mon Aug  1 22:25:58 2011
Return-Path: <david.black@emc.com>
X-Original-To: tsv-area@ietfa.amsl.com
Delivered-To: tsv-area@ietfa.amsl.com
Received: from localhost (localhost [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id 1566821F8E7D; Mon,  1 Aug 2011 22:25:58 -0700 (PDT)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -104.553
X-Spam-Level: 
X-Spam-Status: No, score=-104.553 tagged_above=-999 required=5 tests=[AWL=-1.942, BAYES_00=-2.599, FRT_STOCK2=3.988, RCVD_IN_DNSWL_MED=-4, USER_IN_WHITELIST=-100]
Received: from mail.ietf.org ([64.170.98.30]) by localhost (ietfa.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id 1LI3u30pYkJC; Mon,  1 Aug 2011 22:25:56 -0700 (PDT)
Received: from mexforward.lss.emc.com (mexforward.lss.emc.com [128.222.32.20]) by ietfa.amsl.com (Postfix) with ESMTP id 4384A21F8E7E; Mon,  1 Aug 2011 22:25:47 -0700 (PDT)
Received: from hop04-l1d11-si01.isus.emc.com (HOP04-L1D11-SI01.isus.emc.com [10.254.111.54]) by mexforward.lss.emc.com (Switch-3.4.3/Switch-3.4.3) with ESMTP id p725PkHB007803 (version=TLSv1/SSLv3 cipher=DHE-RSA-AES256-SHA bits=256 verify=NO); Tue, 2 Aug 2011 01:25:47 -0400
Received: from mailhub.lss.emc.com (mailhubhoprd03.lss.emc.com [10.254.221.145]) by hop04-l1d11-si01.isus.emc.com (RSA Interceptor); Tue, 2 Aug 2011 01:25:32 -0400
Received: from mxhub12.corp.emc.com (mxhub12.corp.emc.com [10.254.92.107]) by mailhub.lss.emc.com (Switch-3.4.3/Switch-3.4.3) with ESMTP id p725PUtp022163; Tue, 2 Aug 2011 01:25:30 -0400
Received: from mx14a.corp.emc.com ([169.254.1.245]) by mxhub12.corp.emc.com ([10.254.92.107]) with mapi; Tue, 2 Aug 2011 01:25:29 -0400
From: <david.black@emc.com>
To: <bittau@cs.stanford.edu>
Date: Tue, 2 Aug 2011 01:25:27 -0400
Subject: RE: [saag] fyi: tcpcrypt draft
Thread-Topic: [saag] fyi: tcpcrypt draft
Thread-Index: AcxQedwxl66+yMjXQZiRP9KHfYWF/AAEVrDw
Message-ID: <7C4DFCE962635144B8FAE8CA11D0BF1E0589513E2B@MX14A.corp.emc.com>
References: <4DC6C56D.6040709@ieca.com> <D1CCED3D-C041-4DA2-B999-D3EF7FC889D0@cisco.com> <20110727184256.GA3690@shorty> <8A00B447-53C3-44D0-8238-8DAAE44688CB@mindspring.com> <20110730014544.GM3751@shorty> <4884218B-9477-4F6D-84BB-9EB07F4D5CA9@mindspring.com>
In-Reply-To: <4884218B-9477-4F6D-84BB-9EB07F4D5CA9@mindspring.com>
Accept-Language: en-US
Content-Language: en-US
X-MS-Has-Attach: 
X-MS-TNEF-Correlator: 
acceptlanguage: en-US
Content-Type: text/plain; charset="iso-8859-1"
Content-Transfer-Encoding: quoted-printable
MIME-Version: 1.0
X-EMM-MHVC: 1
Cc: tcpcrypt@scs.stanford.edu, tsv-area@ietf.org, saag@ietf.org
X-BeenThere: tsv-area@ietf.org
X-Mailman-Version: 2.1.12
Precedence: list
List-Id: IETF Transport and Services Area Mailing List <tsv-area.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/options/tsv-area>, <mailto:tsv-area-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/tsv-area>
List-Post: <mailto:tsv-area@ietf.org>
List-Help: <mailto:tsv-area-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/tsv-area>, <mailto:tsv-area-request@ietf.org?subject=subscribe>
X-List-Received-Date: Tue, 02 Aug 2011 05:25:58 -0000

Hi Andrea,

> > tcpcrypt fundamentally does however have lower latency for connection
> > establishment compared to TLS due to fewer RTTs needed in negotiating a
> > key (we leverage the 3-way handshake and add only an extra leg),
> > especially when session caching.

There's a problem here that needs attention.  As was pointed out in the tsv=
area presentation (IIRC) of tcpcrypt, there is plenty of unpleasant deploym=
ent experience with at least middleboxes (e.g., firewalls) that discard TCP=
 packets with unexpected header contents such as a new TCP option (or even =
attempted use of the ECN bits).  Sending the SYN with a new TCP option risk=
s paying a long timeout with a site that doesn't support that new TCP optio=
n and has a middlebox that discards the SYN packet because it doesn't like =
the new option - that behavior will complicate incremental adoption.

What to do about this is a longer discussion, as some of the alternatives h=
ave downsides (e.g., starting both a tcpcyrpt and non-tcpcrypt TCP connecti=
ons consumes an additional connection slot at the responder).

Thanks,
--David
----------------------------------------------------
David L. Black, Distinguished Engineer
EMC Corporation, 176 South St., Hopkinton, MA=A0 01748
+1 (508) 293-7953=A0=A0=A0=A0=A0=A0=A0=A0=A0=A0=A0=A0 FAX: +1 (508) 293-778=
6
david.black@emc.com=A0=A0=A0=A0=A0=A0=A0 Mobile: +1 (978) 394-7754
----------------------------------------------------
> -----Original Message-----
> From: tsv-area-bounces@ietf.org [mailto:tsv-area-bounces@ietf.org] On Beh=
alf Of David A. McGrew
> Sent: Monday, August 01, 2011 1:25 PM
> To: Andrea Bittau
> Cc: tcpcrypt@scs.stanford.edu; tsv-area@ietf.org; saag@ietf.org
> Subject: Re: [saag] fyi: tcpcrypt draft
>=20
> Hi Andrea,
>=20
> thanks for the detail.
>=20
> On Jul 29, 2011, at 6:45 PM, Andrea Bittau wrote:
>=20
> > On Thu, Jul 28, 2011 at 07:33:05AM -0700, David A. McGrew wrote:
> >> A lot of work has been done on using IPsec to solve these problems,
> >> and it is important to explain in detail how tcpcrypt improves on
> >> this work.  Besides the opportunistic encryption effort, there is
> >> BTNS (draft-ietf-btns-channel-binding-api-00 defines channel
> >> bindings for IPsec, for instance), RFC 3947 for running IPsec over
> >> NAT.  (Half joking: would the configuration of
> >> draft-ietf-ipsec-internet-key-00 with NAT-T solve all or most these
> >> problems?)
> >
> > tcpcrypt makes incremental deployment easier because probing is built
> > into existing packets and fallback comes at no extrea cost.   When a
> > SYN
> > packet is sent out and tcpcrypt is not supported at the other end, we
> > can fallback to regular tcpcrypt at no extra cost - just carry on
> > with a
> > vanilla TCP handshake.  How would this work with BTNS or IPSec?  Do
> > you
> > probe for IPSec (and then probe for NAT-T), wait for a timeout, and
> > then
> > send a TCP SYN?  If so, it adds network overhead (extra packets) and
> > latency (probe timeout).
>=20
> Yes, valid points.  It would be good to add this motivating text to
> the draft.
>=20
> >
> > tcpcrypt also has less protocol (space) overhead.  NAT-T requires
> > encapsulating IP in UDP, and NAT seems like a common deployment
> > scenario.  It also requires opening more ports on a firewall and
> > configuring additional services (NAT-T).
> >
> >> What TCP applications can't use TLS?   Do you mean that some apps
> >> don't use TLS in current implementations, or that there is some
> >> basic incompatibility?
> >
> > There's a bootstrap problem.  If I have SMTP on TCP/25 I have to
> > assume
> > it's clear-text (I can't send an SSL handshake packet).  Thankfully,
> > SMTP is an extensible protocol and I can send STARTTLS and then go on
> > with SSL.  If the protocol didn't permit any "options" / commands, I'd
> > be stuck to plain-text.  I'd have to allocate a new port.  That makes
> > incremental deployment harder - do I send two SYNs?
>=20
> Got it.  I would state it this way: while TLS can be used to protect
> any TCP application in principle, in practice it is not always
> available, and introducing it into an application protocol requires a
> mechanism for advertising and selecting to use TLS where it is
> available, and falling back on unprotected communications where it is
> not available.
>=20
> >
> >> What about running TCP on top of DTLS?   That would protect the TCP
> >> session, while keeping the crypto separate from the TCP stack.
> >
> > Same answer as per BTNS and NAT-T - we can efficiently incrementally
> > deploy and have less protocol overhead.
> >
> >> 1.  How easy to configure will tcpcrypt be once it supports mutual
> >> authentication, the enforcement of configurable security policies,
> >> and crypto algorithm agility (e.g. ECDH and ECDSA)?
> >
> > Authentication is left to the application, so tcpcrypt will support
> > authentication only if the application supports it.  If the
> > application
> > supports it, then a mechanism is already in place, e.g., htpasswd in
> > Apache for passwords.  So in the case of Apache, it'll include
> > tcpcrypt's session id when doing http digest authentication, for
> > example.
> >
> > We're also working on an authentication library that attempts to
> > authenticate tcpcrypt using DNSSEC+DANE that can be used by all apps
> > if
> > DANE is deployed.  The configuration will require storing a
> > certificate
> > fingerprint in DNS.
> >
> > We expect that most applications (e.g., the ones running clear-text
> > today) won't care about the crypto algorithms used in tcpcrypt's
> > handshake.  So the security policy will be set system-wide (e.g.,
> > allow
> > only RSA, SHA1 or whatever).  For security concious applications it'll
> > be a setsockopt and presumably those applications already have a
> > configuration file for whatever the already use today for crypto.
> >
> >> 2.  The comparison of tcpcrypt to TLS is only considering the RSA
> >> ciphersuites in the latter protocol.  The efficiency and security
> >> properties of other ciphersuites are quite different.
> >
> > Yes, but we expect tcpcrypt to be no worse (or better) than TLS.
>=20
> Agreed; but note that this is a far different statement than "36 x
> better performance", which if I recall right was stated in the
> presentation at IETF81.  It is important to provide a detailed
> comparison, especially to the non-crypto audience who might miss the
> full implications.
>=20
>=20
> > tcpcrypt fundamentally does however have lower latency for conncection
> > establishment compared to TLS due to fewer RTTs needed in
> > negotiating a
> > key (we leverage the 3-way handshake and add only an extra leg),
> > especially when session caching.
> >
> >> 3.  Reversing the client and server public/private key operations in
> >> tcpcrypt, which is used to make the protocol more scalable, is
> >> applicable only to the RSA algorithm.   In particular, the situation
> >> with elliptic curve cryptography is quite different.  This raises
> >> some questions: how does tcpcrypt with RSA compare to an ECDH/ECDSA
> >> TLS ciphersuite, say, at security levels of 128, 192, or 256 bits?
> >> How does tcpcrypt perform with other algorithms?   Given the
> >> interest in ECC as a future direction for cryptography, these
> >> questions deserve some thought.
> >
> > Yes, we'll implement those but we do not expect tcpcrypt's performance
> > to be much different from a TLS implementation - we both do the same
> > thing (e.g., ECDH).
>=20
> What will be especially interesting is to see how the reversal of RSA
> private/public key operations compares to ECC as a function of
> security level.
>=20
> >
> >> What you are describing here is true for the RSA ciphersuites in
> >> SSL, but is not true for the Diffie-Hellman ciphersuites, or the
> >> ECDH ones (see Section 7.4.2 of RFC 5246 for instance).
> >
> > tcpcrypyt's main goal is to be deployed ubiquitously.  Improved
> > performance in one case is only a bonus.  It has a bunch of other
> > advantages (single integrated mechanism, incrementally deployable)
> > apart
> > from RSA performance.
> >
> >> Why is graceful fallback easier at the TCP layer than at another
> >> layer?   And if there is some mechanism within TCP that facilitates
> >> fallback, would it be possible to use that mechanism in conjunction
> >> with a cryptographic protocol external to TCP?  That last option
> >> would minimize the impact to TCP implementations, meaning less new
> >> code inside kernels, and fewer states, and so on.
> >
> > Gracefully, and efficient, I should add.  The problem is detecting
> > whether the other end supports crypto mechanism X.  At a below-
> > transport
> > layer, we'd have to send our IP packet, wait for a timeout, and then
> > send a SYN.  At a transport layer we're sending out a SYN so we can
> > easily piggyback options (perfect!).  At an application layer what's
> > the
> > first thing we send?  STARTTLS?  Does the application-layer protocol
> > allow such new commands or will it drop the command?
> >
> > The minimum TCP mechanism needed for a graceful and efficient fallback
> > is a TCP option in the SYN.  So, tcpcrypt-lite would modify a kernel
> > with a setsockopt(CRYPTO_OPTION) to include the CRYPTO option in a SYN
> > and getsockopt(CRYPTO_OPTION) to see if the CRYPTO option was echoed
> > in
> > the SYN-ACK.  At that point SSL_accept() can be called.
>=20
> This sounds like an option worth pursuing, no pun intended.  You make
> good points about why TCP is a good place to do the capability
> discovery and fallback for crypto.   Why not define a TCP option that
> can enable devices to discover TLS or tcpcrypt or other higher-layer
> crypto protocols?   That option would be much easier to implement and
> standardize, and it could tie into a full tcpcrypt protocol that
> resides in a shim above the TCP layer.  The separation will create
> useful implementation alternatives, and will isolate the TCP state
> machine from changes.
>=20
> For concreteness, I'm thinking of a TCP option that looks like this:
>=20
>                 +---------+---------+-------------------+
>                 |  Kind   |Length=3D2 |     SECCAP        |
>                 +---------+---------+-------------------+
>=20
> The Security Capabilities (SECCAP) field contains an unsigned integer
> that indicates
> the cryptographic security protocol, if any, the endpoint can use to
> protect subsequent
> communications.
>=20
>    SECCAP Value     Meaning
>=20
>          0          Reserved for future use
>          1          No security protocol is available
>          2          TCPcrypt
>          3          TLS
>          ...        ...
>=20
>=20
>=20
> >  But we lose all
> > TCP header MACing - e.g., someone can RST our connection.  TCP AO
> > suggests that MACing TCP headers is important so crypto protocols
> > should
> > do that (hence why we live in the transport layer).
>=20
> True, but why not design tcpcrypt so that it make use of TCP-AO?
> That seems like an approach that would work with the separation
> outlined above.
>=20
> > Also, we can
> > leverage the existing 3-way handshake to get a head start on crypto
> > negotiations.  Sesison caching completes within the three way
> > handshake.
> >
> >> If the RSA keypair is generated using a pseudorandom process (and
> >> there are standards for doing that), the min entropy could be lower.
> >>
> >> But if it is adequate to have the entropy show up in the nonces, the
> >> min-entropy of the key would be a moot point.
> >
> > The pseudo-random generator has to have enough state to generate prime
> > numbers with min entropy 256.  Depending on the prime number
> > generator,
> > this might, for instance, require something on the order of 256 + ceil
> > (log_2 (log_e (2 ^ 1024))) or 266 bits.  We assume most RSA
> > implementations would have more random state than that.
>=20
> Unfortunately that is not always a safe assumption.  There are
> standards for RSA key generation (see for instance NIST SP 800 56-B
> Section 6.3.1) in which n bits of random data are used to seed a
> pseudorandom process that selects the primes used as the private key;
> for 1024-bit RSA, n will be about 80.
>=20
> I am confident that tcpcrypt can be made to work around this fact, but
> it may be an important detail (in the security proof if not in
> practice).
>=20
> David
>=20
> > _______________________________________________
> > saag mailing list
> > saag@ietf.org
> > https://www.ietf.org/mailman/listinfo/saag
>=20


From Michael.Scharf@alcatel-lucent.com  Tue Aug  2 08:40:20 2011
Return-Path: <Michael.Scharf@alcatel-lucent.com>
X-Original-To: tsv-area@ietfa.amsl.com
Delivered-To: tsv-area@ietfa.amsl.com
Received: from localhost (localhost [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id 6717D21F85AE; Tue,  2 Aug 2011 08:40:20 -0700 (PDT)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -6.219
X-Spam-Level: 
X-Spam-Status: No, score=-6.219 tagged_above=-999 required=5 tests=[AWL=0.030,  BAYES_00=-2.599, HELO_EQ_DE=0.35, RCVD_IN_DNSWL_MED=-4]
Received: from mail.ietf.org ([64.170.98.30]) by localhost (ietfa.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id ul-1i1Y1jF8q; Tue,  2 Aug 2011 08:40:19 -0700 (PDT)
Received: from mailrelay2.alcatel.de (mailrelay2.alcatel.de [194.113.59.96]) by ietfa.amsl.com (Postfix) with ESMTP id 8C14621F858D; Tue,  2 Aug 2011 08:40:19 -0700 (PDT)
Received: from SLFSNX.rcs.alcatel-research.de (slfsn1.rcs.de.alcatel-lucent.com [149.204.60.98]) by mailrelay2.alcatel.de (8.14.3/8.14.3/ICT) with ESMTP id p72FeJ9d023709; Tue, 2 Aug 2011 17:40:19 +0200
X-MimeOLE: Produced By Microsoft Exchange V6.5
Content-class: urn:content-classes:message
MIME-Version: 1.0
Content-Type: text/plain; charset="us-ascii"
Content-Transfer-Encoding: quoted-printable
Subject: RE: [saag] fyi: tcpcrypt draft
Date: Tue, 2 Aug 2011 17:40:18 +0200
Message-ID: <133D9897FB9C5E4E9DF2779DC91E947C0654BD89@SLFSNX.rcs.alcatel-research.de>
In-Reply-To: <20110801191407.GA3765@shorty>
X-MS-Has-Attach: 
X-MS-TNEF-Correlator: 
Thread-Topic: [saag] fyi: tcpcrypt draft
Thread-Index: AcxQf3abwAXl+QxvSQyWhoUPGc+SjwAgsV6w
References: <4DC6C56D.6040709@ieca.com><D1CCED3D-C041-4DA2-B999-D3EF7FC889D0@cisco.com><20110727184256.GA3690@shorty><8A00B447-53C3-44D0-8238-8DAAE44688CB@mindspring.com><20110730014544.GM3751@shorty><4884218B-9477-4F6D-84BB-9EB07F4D5CA9@mindspring.com> <20110801191407.GA3765@shorty>
From: "SCHARF, Michael" <Michael.Scharf@alcatel-lucent.com>
To: "Andrea Bittau" <bittau@cs.stanford.edu>, "David A. McGrew" <david.a.mcgrew@mindspring.com>
X-Scanned-By: MIMEDefang 2.64 on 149.204.45.73
Cc: tcpcrypt@scs.stanford.edu, tsv-area@ietf.org, saag@ietf.org
X-BeenThere: tsv-area@ietf.org
X-Mailman-Version: 2.1.12
Precedence: list
List-Id: IETF Transport and Services Area Mailing List <tsv-area.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/options/tsv-area>, <mailto:tsv-area-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/tsv-area>
List-Post: <mailto:tsv-area@ietf.org>
List-Help: <mailto:tsv-area-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/tsv-area>, <mailto:tsv-area-request@ietf.org?subject=subscribe>
X-List-Received-Date: Tue, 02 Aug 2011 15:40:20 -0000

> On Mon, Aug 01, 2011 at 10:25:14AM -0700, David A. McGrew wrote:
> > This sounds like an option worth pursuing, no pun intended.=20
>  You make=20
> > good points about why TCP is a good place to do the capability
> > discovery and fallback for crypto.   Why not define a TCP option
> > that can enable devices to discover TLS or tcpcrypt or other
> > higher-layer crypto protocols?   That option would be much easier to
> > implement and standardize, and it could tie into a full tcpcrypt=20
> > protocol that resides in a shim above the TCP layer.  The=20
> separation=20
> > will create useful implementation alternatives, and will=20
> isolate the=20
> > TCP state machine from changes.
>=20
> We have been considering this option throughout the work (and=20
> will still
> do) but:
>=20
> * No TCP header protection.  TCP-AO, which you suggest, does not
>   interoperate with NATs.  We want something that works "out of the
>   box".
>=20
> * We can't leverage the 3-way handshake if operating at the=20
> application
>   layer.  With tcpcrypt, we can do all session key negotiation in a
>   4-way handshake (just one extra leg), or 3 if session=20
> caching.  Saving
>   RTTs seems like a good thing.
>=20
> From a clean-slate perspective, it seems that integrating=20
> this into TCP has all the advantages we're looking for. =20
> Doing minimal work at TCP (like a signle "do you crypto?"=20
> option) and doing the rest at an application layer seems to=20
> come at a compromise (e.g., the drawbacks listed above).
>=20
> From an implementation perspective, I agree that the least=20
> changes to TCP are probably the best option.  tcpcrypt, at=20
> least for us, seems to have a good balance between=20
> clean-slate and (relatively) few changes to TCP, hence why=20
> we're pushing for it, as is.  If there were=20
> non-implementation arguments as to app vs. transport layer=20
> we'd love to hear those.

For whatever it is worth, the question of TCP extension vs. shim layer
seems to be somehow related to some discussions in the MPTCP WG last
year. In a different context, an expired ID
(http://tools.ietf.org/html/draft-scharf-mptcp-mctcp-01) discusses some
pros and cons of transport layer vs. shim layer+discovery by SYN
options. Not all of them are about implementation, and IMHO both
alternatives have disadvantages. Section 5.4 of RFC 6182 also includes
some thoughts.

Michael

From bittau@gmail.com  Tue Aug  2 14:20:03 2011
Return-Path: <bittau@gmail.com>
X-Original-To: tsv-area@ietfa.amsl.com
Delivered-To: tsv-area@ietfa.amsl.com
Received: from localhost (localhost [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id 43DF311E80A9; Tue,  2 Aug 2011 14:20:03 -0700 (PDT)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -3.092
X-Spam-Level: 
X-Spam-Status: No, score=-3.092 tagged_above=-999 required=5 tests=[AWL=0.507,  BAYES_00=-2.599, RCVD_IN_DNSWL_LOW=-1]
Received: from mail.ietf.org ([64.170.98.30]) by localhost (ietfa.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id nIExaV82IBGS; Tue,  2 Aug 2011 14:20:02 -0700 (PDT)
Received: from mail-yi0-f44.google.com (mail-yi0-f44.google.com [209.85.218.44]) by ietfa.amsl.com (Postfix) with ESMTP id A5D0511E808F; Tue,  2 Aug 2011 14:19:27 -0700 (PDT)
Received: by yie30 with SMTP id 30so140597yie.31 for <multiple recipients>; Tue, 02 Aug 2011 14:19:37 -0700 (PDT)
DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=gmail.com; s=gamma; h=sender:date:from:to:cc:subject:message-id:references:mime-version :content-type:content-disposition:in-reply-to:x-echelon:user-agent; bh=LDUlszumrTIK5f+7RhdnXO39eiABbYTD6rnGpsLLpZ4=; b=PmRR7HLYVqEl52FTVYcZcdK3ydfEVSpq0UU4H2NMR6vBFybzDW085mOx+QXnoGzjtM octgPn0fOKgMYjYsepbXNGbS9PckSkMbkSxp8ovRET1NgNpeqWw/ZFFvfNBz6X36oq+D 24VlrcmXPZ6UTcWBf0pO/73gG3tEmpDVSPGtY=
Received: by 10.68.31.1 with SMTP id w1mr11221021pbh.421.1312319976916; Tue, 02 Aug 2011 14:19:36 -0700 (PDT)
Received: from tapir.cs.ucl.ac.uk (sorbox.scs.stanford.edu [171.66.3.197]) by mx.google.com with ESMTPS id k3sm222187pbj.45.2011.08.02.14.19.33 (version=TLSv1/SSLv3 cipher=OTHER); Tue, 02 Aug 2011 14:19:34 -0700 (PDT)
Sender: Andrea Bittau <bittau@gmail.com>
Received: by tapir.cs.ucl.ac.uk (Postfix, from userid 0) id 181524810171; Tue,  2 Aug 2011 14:17:54 -0700 (PDT)
Date: Tue, 2 Aug 2011 14:17:54 -0700
From: Andrea Bittau <bittau@cs.stanford.edu>
To: david.black@emc.com
Subject: Re: [saag] fyi: tcpcrypt draft
Message-ID: <20110802211753.GI3814@shorty>
References: <4DC6C56D.6040709@ieca.com> <D1CCED3D-C041-4DA2-B999-D3EF7FC889D0@cisco.com> <20110727184256.GA3690@shorty> <8A00B447-53C3-44D0-8238-8DAAE44688CB@mindspring.com> <20110730014544.GM3751@shorty> <4884218B-9477-4F6D-84BB-9EB07F4D5CA9@mindspring.com> <7C4DFCE962635144B8FAE8CA11D0BF1E0589513E2B@MX14A.corp.emc.com>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <7C4DFCE962635144B8FAE8CA11D0BF1E0589513E2B@MX14A.corp.emc.com>
X-Echelon: Bush Bomb War KGB
User-Agent: Mutt/1.5.21 (2010-09-15)
Cc: tcpcrypt@scs.stanford.edu, tsv-area@ietf.org, saag@ietf.org
X-BeenThere: tsv-area@ietf.org
X-Mailman-Version: 2.1.12
Precedence: list
List-Id: IETF Transport and Services Area Mailing List <tsv-area.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/options/tsv-area>, <mailto:tsv-area-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/tsv-area>
List-Post: <mailto:tsv-area@ietf.org>
List-Help: <mailto:tsv-area-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/tsv-area>, <mailto:tsv-area-request@ietf.org?subject=subscribe>
X-List-Received-Date: Tue, 02 Aug 2011 21:20:03 -0000

On Tue, Aug 02, 2011 at 01:25:27AM -0400, david.black@emc.com wrote:
> There's a problem here that needs attention.  As was pointed out in the tsvarea presentation (IIRC) of tcpcrypt, there is plenty of unpleasant deployment experience with at least middleboxes (e.g., firewalls) that discard TCP packets with unexpected header contents such as a new TCP option (or even attempted use of the ECN bits).  Sending the SYN with a new TCP option risks paying a long timeout with a site that doesn't support that new TCP option and has a middlebox that discards the SYN packet because it doesn't like the new option - that behavior will complicate incremental adoption.
> 
> What to do about this is a longer discussion, as some of the alternatives have downsides (e.g., starting both a tcpcyrpt and non-tcpcrypt TCP connections consumes an additional connection slot at the responder).

Yes, that's the crux of the argument - will a new TCP option cause
packets to be dropped?  In fact, that's the only compelling argument for
why an application layer approach might be better than a TCP one, though
we still think the arguments for TCP outweigh those for the application
layer.

We're working on measuring (the best we can) how much middleboxes drop
packets with unknown options to see whether it's a show stopper.
Hopefully the common case will be that middleboxes strip options rather
than dropping packets.  In fact, if you have pointers to case studies on
middlebox behavior we'd love to have those.

As you say the solution is not straightforward.  For example, a very
conservative approach could be to disable tcpcrypt if the sender is
unable to connect using tcpcrypt to (say) three sites.  That way the
penatly of sending a vanilla TCP syn and a tcpcrypt SYN to "path probe"
is paid only three times.  tcpcrypt can be turned back on when the
sender's IP changes.  This may work out pretty well if our test results
show that middleboxes are most often found closer to the client-side
(e.g., NATs) rather than the server side.

My understanding of most of ECN's problems is that the bits were
previosuly reserved and then overloaded.  TCP options (in principle) are
a by-the-spec way of extending TCP, so hopefully we'll suffer less
problems compared to ECN. 

From bittau@gmail.com  Tue Aug  2 14:32:34 2011
Return-Path: <bittau@gmail.com>
X-Original-To: tsv-area@ietfa.amsl.com
Delivered-To: tsv-area@ietfa.amsl.com
Received: from localhost (localhost [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id BA2DB11E80A9; Tue,  2 Aug 2011 14:32:34 -0700 (PDT)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -3.165
X-Spam-Level: 
X-Spam-Status: No, score=-3.165 tagged_above=-999 required=5 tests=[AWL=0.434,  BAYES_00=-2.599, RCVD_IN_DNSWL_LOW=-1]
Received: from mail.ietf.org ([64.170.98.30]) by localhost (ietfa.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id bLxHltag1wfg; Tue,  2 Aug 2011 14:32:34 -0700 (PDT)
Received: from mail-pz0-f53.google.com (mail-pz0-f53.google.com [209.85.210.53]) by ietfa.amsl.com (Postfix) with ESMTP id 18D4411E808F; Tue,  2 Aug 2011 14:32:34 -0700 (PDT)
Received: by pzk6 with SMTP id 6so281580pzk.26 for <multiple recipients>; Tue, 02 Aug 2011 14:32:44 -0700 (PDT)
DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=gmail.com; s=gamma; h=sender:date:from:to:cc:subject:message-id:references:mime-version :content-type:content-disposition:in-reply-to:x-echelon:user-agent; bh=SZKBd8x9gTlFFgBu1r8dofaT2HwXSebZnuuGSI1/QYM=; b=fgX06X3COTwUj3qHmvsMio10g65Sezbky48UU7Ttmsxs7DngmYrqQASnRaKC9alFDD Ra+RKH/gUDV+a3NQeyD59uaOeLHRdjTdBop0PiQiWrs2n76MzknduzkVGs6DzI3OHoW9 ZSijrcu4grBd++06P8fZjCvcGXqPrF45yvrjI=
Received: by 10.68.13.38 with SMTP id e6mr6239984pbc.72.1312320764132; Tue, 02 Aug 2011 14:32:44 -0700 (PDT)
Received: from tapir.cs.ucl.ac.uk (sorbox.scs.stanford.edu [171.66.3.197]) by mx.google.com with ESMTPS id l7sm233186pbh.42.2011.08.02.14.32.41 (version=TLSv1/SSLv3 cipher=OTHER); Tue, 02 Aug 2011 14:32:42 -0700 (PDT)
Sender: Andrea Bittau <bittau@gmail.com>
Received: by tapir.cs.ucl.ac.uk (Postfix, from userid 0) id A282F4810171; Tue,  2 Aug 2011 14:31:01 -0700 (PDT)
Date: Tue, 2 Aug 2011 14:31:01 -0700
From: Andrea Bittau <bittau@cs.stanford.edu>
To: "SCHARF, Michael" <Michael.Scharf@alcatel-lucent.com>
Subject: Re: [saag] fyi: tcpcrypt draft
Message-ID: <20110802213101.GJ3814@shorty>
References: <4DC6C56D.6040709@ieca.com> <D1CCED3D-C041-4DA2-B999-D3EF7FC889D0@cisco.com> <20110727184256.GA3690@shorty> <8A00B447-53C3-44D0-8238-8DAAE44688CB@mindspring.com> <20110730014544.GM3751@shorty> <4884218B-9477-4F6D-84BB-9EB07F4D5CA9@mindspring.com> <20110801191407.GA3765@shorty> <133D9897FB9C5E4E9DF2779DC91E947C0654BD89@SLFSNX.rcs.alcatel-research.de>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <133D9897FB9C5E4E9DF2779DC91E947C0654BD89@SLFSNX.rcs.alcatel-research.de>
X-Echelon: Bush Bomb War KGB
User-Agent: Mutt/1.5.21 (2010-09-15)
Cc: tcpcrypt@scs.stanford.edu, tsv-area@ietf.org, "David A. McGrew" <david.a.mcgrew@mindspring.com>, saag@ietf.org
X-BeenThere: tsv-area@ietf.org
X-Mailman-Version: 2.1.12
Precedence: list
List-Id: IETF Transport and Services Area Mailing List <tsv-area.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/options/tsv-area>, <mailto:tsv-area-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/tsv-area>
List-Post: <mailto:tsv-area@ietf.org>
List-Help: <mailto:tsv-area-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/tsv-area>, <mailto:tsv-area-request@ietf.org?subject=subscribe>
X-List-Received-Date: Tue, 02 Aug 2011 21:32:34 -0000

On Tue, Aug 02, 2011 at 05:40:18PM +0200, SCHARF, Michael wrote:
> For whatever it is worth, the question of TCP extension vs. shim layer
> seems to be somehow related to some discussions in the MPTCP WG last
> year. In a different context, an expired ID
> (http://tools.ietf.org/html/draft-scharf-mptcp-mctcp-01) discusses some
> pros and cons of transport layer vs. shim layer+discovery by SYN
> options. Not all of them are about implementation, and IMHO both
> alternatives have disadvantages. Section 5.4 of RFC 6182 also includes
> some thoughts.

Mark Handley is in fact one of the authors of tcpcrypt and RFC 6182 so
we're working closely with him to figure out what the implications of
TCP options are in practice.  Section 5.4 of RFC 6182 concludes that TCP
options are the right way to go and hopefully that applies to tcpcrypt
too (we're working on it with Mark).  All tcpcrypt authors agree that
TCP is the right place for encryption from a design perspective, though
the one practical downside are middleboxes.  We're hoping we can
"handle" most cases by downgrading to TCP by checking if the options
make it through on the SYN / SYN-ACK.  A pretty bad case is when the SYN
is dropped.

For us to better present an argument (or perhaps even change our
design!) we're trying to figure out what the arguments are for an
application layer approach as opposed to a transport one.
Implementation has already come up.  From a design perspective however,
apart from middleboxes (or even more specifically, SYNs getting dropped)
we'd be glad to hear more opinions on why the application layer might be
more attractive.

From c.raiciu@cs.ucl.ac.uk  Wed Aug  3 04:44:52 2011
Return-Path: <c.raiciu@cs.ucl.ac.uk>
X-Original-To: tsv-area@ietfa.amsl.com
Delivered-To: tsv-area@ietfa.amsl.com
Received: from localhost (localhost [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id 6FD5B21F8B21; Wed,  3 Aug 2011 04:44:52 -0700 (PDT)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -2.599
X-Spam-Level: 
X-Spam-Status: No, score=-2.599 tagged_above=-999 required=5 tests=[BAYES_00=-2.599]
Received: from mail.ietf.org ([64.170.98.30]) by localhost (ietfa.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id CP6UD+QFbLyQ; Wed,  3 Aug 2011 04:44:51 -0700 (PDT)
Received: from bells2.cs.ucl.ac.uk (bells2.cs.ucl.ac.uk [128.16.5.33]) by ietfa.amsl.com (Postfix) with ESMTP id B5EA421F8B18; Wed,  3 Aug 2011 04:44:51 -0700 (PDT)
Received: from [141.85.37.138] (helo=[10.0.0.3]) by bells2.cs.ucl.ac.uk with esmtpsa (TLSv1:AES128-SHA:128) (C.Raiciu authenticated) (Exim 4.54) id 1QoZrV-0009X5-Cm; Wed, 03 Aug 2011 12:43:45 +0100
Subject: Re: [saag] fyi: tcpcrypt draft
Mime-Version: 1.0 (Apple Message framework v1081)
Content-Type: text/plain; charset=us-ascii
From: Costin Raiciu <c.raiciu@cs.ucl.ac.uk>
In-Reply-To: <20110802213101.GJ3814@shorty>
Date: Wed, 3 Aug 2011 14:44:50 +0300
Content-Transfer-Encoding: quoted-printable
Message-Id: <9AE53878-7A53-4AD2-80C2-07ADD000E479@cs.ucl.ac.uk>
References: <4DC6C56D.6040709@ieca.com> <D1CCED3D-C041-4DA2-B999-D3EF7FC889D0@cisco.com> <20110727184256.GA3690@shorty> <8A00B447-53C3-44D0-8238-8DAAE44688CB@mindspring.com> <20110730014544.GM3751@shorty> <4884218B-9477-4F6D-84BB-9EB07F4D5CA9@mindspring.com> <20110801191407.GA3765@shorty> <133D9897FB9C5E4E9DF2779DC91E947C0654BD89@SLFSNX.rcs.alcatel-research.de> <20110802213101.GJ3814@shorty>
To: Andrea Bittau <bittau@cs.stanford.edu>
X-Mailer: Apple Mail (2.1081)
Cc: tcpcrypt@scs.stanford.edu, tsv-area@ietf.org, "David A. McGrew" <david.a.mcgrew@mindspring.com>, saag@ietf.org
X-BeenThere: tsv-area@ietf.org
X-Mailman-Version: 2.1.12
Precedence: list
List-Id: IETF Transport and Services Area Mailing List <tsv-area.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/options/tsv-area>, <mailto:tsv-area-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/tsv-area>
List-Post: <mailto:tsv-area@ietf.org>
List-Help: <mailto:tsv-area-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/tsv-area>, <mailto:tsv-area-request@ietf.org?subject=subscribe>
X-List-Received-Date: Wed, 03 Aug 2011 11:44:52 -0000

Hi,

Michio Honda has recently performed experiments on exactly these issues =
of extending TCP; the results are reported in an IMC paper to be =
published later this year that Michio and us have co-authored.

I can send the paper to interested people (please email me directly) but =
can't send it on this list because it is still not in its final form.

The paper probes 150 access networks (including home, hotspots, 3g =
providers, cable, etc) to check and see if new  TCP options get through, =
and whether middleboxes resegment the TCP stream.=20
There are many other tests there, but those are not necesarily relevant =
to tcpcrypt. The paper does not probe the server end of things where =
firewalls, load balancers and the like can cause further  grief.

The new options get through on SYNs on 86% - 96% of the paths we probed. =
The percentage depends on the TCP port (only 86% allow new options on =
HTTP traffic, 94% on HTTPS traffic, and 96% for a new, unknown port).
SYNs with new options were never dropped; in the remaining 4%-14% of the =
cases the new options were simply removed.

As for middleboxes that resegment the bytestream: this behaviour did =
appear in the tests, but it was correlated with the removal of new =
options on SYNs. Basically, middleboxes that resegment tend to terminate =
the connection and open a new one to the destination, acting like a full =
proxy.

We've analyzed all experimental data and its implications for about =
mptcp and tcpcrypt: the result is that these two protocols seem safe to =
deploy today in the access networks we've probed.=20

Cheers,
Costin

On 3 Aug 2011, at 00:31, Andrea Bittau wrote:

> On Tue, Aug 02, 2011 at 05:40:18PM +0200, SCHARF, Michael wrote:
>> For whatever it is worth, the question of TCP extension vs. shim =
layer
>> seems to be somehow related to some discussions in the MPTCP WG last
>> year. In a different context, an expired ID
>> (http://tools.ietf.org/html/draft-scharf-mptcp-mctcp-01) discusses =
some
>> pros and cons of transport layer vs. shim layer+discovery by SYN
>> options. Not all of them are about implementation, and IMHO both
>> alternatives have disadvantages. Section 5.4 of RFC 6182 also =
includes
>> some thoughts.
>=20
> Mark Handley is in fact one of the authors of tcpcrypt and RFC 6182 so
> we're working closely with him to figure out what the implications of
> TCP options are in practice.  Section 5.4 of RFC 6182 concludes that =
TCP
> options are the right way to go and hopefully that applies to tcpcrypt
> too (we're working on it with Mark).  All tcpcrypt authors agree that
> TCP is the right place for encryption from a design perspective, =
though
> the one practical downside are middleboxes.  We're hoping we can
> "handle" most cases by downgrading to TCP by checking if the options
> make it through on the SYN / SYN-ACK.  A pretty bad case is when the =
SYN
> is dropped.
>=20
> For us to better present an argument (or perhaps even change our
> design!) we're trying to figure out what the arguments are for an
> application layer approach as opposed to a transport one.
> Implementation has already come up.  =46rom a design perspective =
however,
> apart from middleboxes (or even more specifically, SYNs getting =
dropped)
> we'd be glad to hear more opinions on why the application layer might =
be
> more attractive.


From ycai@cisco.com  Tue Aug  9 13:49:49 2011
Return-Path: <ycai@cisco.com>
X-Original-To: tsv-area@ietfa.amsl.com
Delivered-To: tsv-area@ietfa.amsl.com
Received: from localhost (localhost [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id 4CF6E21F8B04; Tue,  9 Aug 2011 13:49:49 -0700 (PDT)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -0.532
X-Spam-Level: 
X-Spam-Status: No, score=-0.532 tagged_above=-999 required=5 tests=[BAYES_00=-2.599, RCVD_NUMERIC_HELO=2.067]
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 BBtadf2gVD0A; Tue,  9 Aug 2011 13:49:48 -0700 (PDT)
Received: from rcdn-iport-8.cisco.com (rcdn-iport-8.cisco.com [173.37.86.79]) by ietfa.amsl.com (Postfix) with ESMTP id DA42821F8B02; Tue,  9 Aug 2011 13:49:47 -0700 (PDT)
DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/simple; d=cisco.com; i=ycai@cisco.com; l=10968; q=dns/txt; s=iport; t=1312923018; x=1314132618; h=date:subject:from:to:cc:message-id:in-reply-to: mime-version:content-transfer-encoding; bh=CKrLUywYoxwv2VZEj3FNiVTIxKhr3CKT4sw6BlHN+uY=; b=EOCmw3d7tH3DzQy/0WeQJwSSPGPruhOAqYcjZFGVPAEHhbNmN/+49JVt gPI3UWQfeCAqmyYmM9mVObEZjsYYn7g1D5D9I2w2/OA36lNEz+TBQQzvp FZ7sPwPW0QV/7W07Kukkx55hC/+Mx2tv4vgpVK89yvYr1GsTMxF4Jlvsl c=;
X-IronPort-Anti-Spam-Filtered: true
X-IronPort-Anti-Spam-Result: Ah8IABOdQU6rRDoJ/2dsb2JhbABCpz0Cd4FAAQEBAQIBAQEBDwEnAgExCwUNAQhnBjABAQQBDQUbB4dLBJ8QAZ5ahkYEh1yLKYUShF2DAIQc
X-IronPort-AV: E=Sophos;i="4.67,345,1309737600"; d="scan'208";a="11479663"
Received: from mtv-core-4.cisco.com ([171.68.58.9]) by rcdn-iport-8.cisco.com with ESMTP; 09 Aug 2011 20:50:16 +0000
Received: from xbh-sjc-221.amer.cisco.com (xbh-sjc-221.cisco.com [128.107.191.63]) by mtv-core-4.cisco.com (8.14.3/8.14.3) with ESMTP id p79KoGCj030223; Tue, 9 Aug 2011 20:50:16 GMT
Received: from xmb-sjc-216.amer.cisco.com ([171.70.151.184]) by xbh-sjc-221.amer.cisco.com with Microsoft SMTPSVC(6.0.3790.4675);  Tue, 9 Aug 2011 13:50:16 -0700
Received: from 128.107.161.230 ([128.107.161.230]) by xmb-sjc-216.amer.cisco.com ([171.70.151.184]) with Microsoft Exchange Server HTTP-DAV ; Tue,  9 Aug 2011 20:50:15 +0000
User-Agent: Microsoft-Entourage/12.26.0.100708
Date: Tue, 09 Aug 2011 13:50:21 -0700
Subject: Re: [pim] tsv-dir review of draft-ietf-pim-mtid-08
From: Yiqun Cai <ycai@cisco.com>
To: Marshall Eubanks <marshall.eubanks@gmail.com>, IETF Discussion <ietf@ietf.org>
Message-ID: <CA66EB9D.AF52C%ycai@cisco.com>
Thread-Topic: [pim] tsv-dir review of draft-ietf-pim-mtid-08
Thread-Index: AcxW1fQVCxnlCmf+6kW+cOAFGDDVgQ==
In-Reply-To: <A9EFAA60-85DE-40F1-86C3-B9652E9CC0EA@gmail.com>
Mime-version: 1.0
Content-type: text/plain; charset="US-ASCII"
Content-transfer-encoding: 7bit
X-OriginalArrivalTime: 09 Aug 2011 20:50:16.0307 (UTC) FILETIME=[F149CC30:01CC56D5]
X-Mailman-Approved-At: Wed, 10 Aug 2011 07:13:03 -0700
Cc: TSV Area <tsv-area@ietf.org>, pim@ietf.org, Transport Directorate <tsv-dir@ietf.org>
X-BeenThere: tsv-area@ietf.org
X-Mailman-Version: 2.1.12
Precedence: list
List-Id: IETF Transport and Services Area Mailing List <tsv-area.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/options/tsv-area>, <mailto:tsv-area-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/tsv-area>
List-Post: <mailto:tsv-area@ietf.org>
List-Help: <mailto:tsv-area-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/tsv-area>, <mailto:tsv-area-request@ietf.org?subject=subscribe>
X-List-Received-Date: Tue, 09 Aug 2011 20:49:49 -0000

Marshall,  thank you very much for the review.

Some of my comments are inline.


On 6/30/11 7:22 AM, "Marshall Eubanks" <marshall.eubanks@gmail.com> wrote:

> I've reviewed this document as part of the transport area directorate's
> ongoing effort to review key IETF documents. These comments were written
> primarily for the transport area directors, but are copied to the document's
> authors for their information and to allow them to address any issues raised.
> The authors should consider this review together with any other last-call
> comments they receive. Please always CC tsv-dir@ietf.org if you reply to or
> forward this review.
> 
> Note : I also reviewed the WG mailing list discussions for this document.
> 
> This draft is basically ready for publication, but has some issues that in my
> opinion should be fixed before publication.
> 
> There are, however, some issues that I think that should be addressed, either
> in a follow-on document, or a revision of this document.
> 
> draft-ietf-pim-mtid  introduces a new type of PIM Join Attribute [RFC5384],
> the MT-ID Join Attribute, that extends
> PIM signaling to cover multiple topologies and to enable the identification of
> which topology should be used when
> constructing a particular multicast distribution tree.
> 
> Multiple topologies have been used for some time in unicast (they are
> supported by IS-IS and OSPF), and also in multicast. The most common multicast
> requirement for multiple topologies is for video transport, where important or
> expensive video streams (say, of sporting events) are protected from
> disruption by multiple transport on completely different network paths. In
> that use, if a network link goes down or becomes congested or otherwise
> suboptimal, another source of the video data is already present and the
> display can be seamlessly switched to the other copy of the stream. This use
> case is alluded to in the document.
> 
> Another use case for multiple-topologies is to separate out different types of
> traffic (say, latency sensitive traffic from larger, bulk, flows). This is not
> alluded to in this document, and it is not clear to what extent this was
> intended. These multiple topologies may or may not result
> in the replication of data, and so may not be intended by the author. However,
> I would say that this could be used to support this, and so it will be. If
> there is some reason NOT to use this mechanism for this purpose, some
> description of the reasons why would be in order.
> 
> This document sets up a numerical variable, the  MT-ID Join Attribute, to
> assign to multicast topologies, to be used to separate different topologies
> for different paths. This number for a given multicast topology need not be
> the same as any unicast multiple topology identifiers, and similarly the
> multicast topologies and the underlying unicast topologies may be different.

You are right that there are indeed more use cases that can make use of the
functions described by this draft.  And some do exist in the field.

However, we (as co-authors) were very reluctant to list them.  We felt this
would be more appropriate to be described and published as whitepapers by
vendors, and we indeed have discussed them in public on behalf of our
employer.

It was our intention to try to keep this draft focused on the changes
required from a PIM protocol point of view.  I hope you will understand and
appreciate our viewpoints.


> 
> This document is sort, straight-forward, and relatively well-written. I had a
> few non-fatal issues.  (I would be glad to suggest some text for any of these
> issues to the authors if they interested.)
> 
> Minor Issues :
> 
> Section 3.2 
> 
> "- As shown earlier, this value is not required to be the same as the MT-ID
> used by the unicast routing protocols that contribute routes to the topology.
> In practice, when only one unicast routing protocol (such as OSPF or IS-IS) is
> used, the PIM MT-ID is RECOMMENDED to be assigned using the same value as the
> IGP topology identifier. Using the same example presented earlier, if
> every route in PIM 500 is contributed by OSPF 1000, it is RECOMMENDED to name
> this RPF topology as PIM 1000 instead of PIM 500. This is for the purpose of
> reducing management overhead and simplifying troubleshooting."
> 
> In the actual example, PIM 500 is not the same network topology as OSPF 1000
> (it has an extra link, to the source, obtained by MBGP). This is specifically
> mentioned previously in the text. So, this is NOT the same example as
> presented earlier and it is NOT clear that this recommendation applies here.

What we tried to say here is that,

1. If a PIM topology has routes contributed by different protocols, say OSPF
and MBGP,  it is better to use a different MT-ID for PIM.  For example, use
500 for PIM and 1000 for OSPF.

2. But if a PIM topology has all its routes contributed by a single
protocol,  then it is better to use the same MT-ID as the underlying unicast
protocol and in this case, 1000 is used for both PIM and OSPF.

Please let us know if the explanation helps and we will definitely
appreciate if you suggest some text here.

> 
> If this is left as is, it will certainly confuse some people. I would suggest
> adding text to clarify this, or creating another example where the
> unicast and multicast topologies are the same.
> 
> Section 4.2.3 
> 
> "- There is at most 1 PIM MT-ID attribute encoded. If there are multiple PIM
> MT-ID Join Attributes included, only the last one is accepted for this
> particular source. Processing of the rest of the Join message continues."
> 
> Is this a typo ? 
> 
> s/at most/at least/
> 
> would seem to be appropriate. (It says, "at most" 1, and then describes what
> do to if there is more than 1.) Otherwise, please explain what is meant.

We did mean there can be "at most" 1 MT-ID attribute.

But if there are more than 1 included, most likely due to an implementation
error, the rest of the paragraph suggests what an implementation should do.
Would it be better if we add "due to an implementation error"?

> 
> Section 6.
> 
> The Labels for Section 6.1 and 6.2 are the same. I suspect this is just a
> typo, and they should be
> 
> 6.1. PIM-Hello Options
> 
> 6.2. PIM Join Attribute Types
> 

Good catch.  Will fix.

> More Substantive Issues :
> 
> I had two substantive issues with this document.
> 
> 1.) The range for the topologies. It is highly likely that the users of this
> capability would want to use multiple topologies differently for different
> multicast address ranges. For example, use of two topologies is likely to be
> an expensive customer option, that not all customers would pay for, and so
> would not apply to all multicast addresses. Or, the backup typologies may
> (probably will be) tailored differently for different customers in different
> locations. Or, topologies might be tailored for different data types
> (different backups for real time video conferencing versus TV broadcast
> transmissions, say). This is very briefly alluded to in Section 3.1, but no
> consideration is given to issues arising from this.
> 
> For example, suppose I have two customers, and I wish to have one basic
> topology (say, PIM 100) for 239.10/16, and I wish to have two backup
> topologies for the two customers (say, PIM 1001 for 239.10.1/24 and PIM 1002
> for 239.10.2/24.) Can I have PIM 100 cover both the PIM 1001 and PIM 1002
> range ? Are there issues with this ? Would it be better to split PIM 100 into
> pieces to match 1001 and 1002 ? Nothing like this is discussed

The draft allows the above to happen.  Because as I indicated earlier, the
draft tries to focus on the protocol changes required to support the
functionality but not the deployment cases that can make use of it.

> 
> I can easily see interoperability issues arising from this (i.e., vendor A
> might allow PIM 100 to overlap PIM 1001 and PIM 1002, and vendor B might not).
> I think that some text to discuss the entire issue of ranging (presumably,
> that they can be allowed and overlaps SHOULD be supported) is needed.
> 
> Again, this could be out of scope here, but, then, where will it appear ?

I agree it is out of scope for the reasons I explained earlier but I don't
have a good answer what is the best practice to document all these cases.

> 
> 2.) There is very little discussion as to how to use this capability. Now,
> this might be considered out of scope for this document, but there will be
> issues, and I do not see a follow-on in the pipeline.
> 
> Issues include sharing of links and the general connections between
> topologies. It is likely that in real uses, a portion of the multiple
> topologies may be either physically shared or directly linked. It may not be
> possible to engineer out all single points of failure, for example an undersea
> cable. This is possible with the first use case (independent topologies for
> video) and likely with the second (different topologies for different types of
> content). 
> 

Yes, same as above.

> Issues I see here include
> 
> - the actual multicast data will not include any PIM attributes.  How, then,
> is a midpoint router receiving two copies of every packet know that it
> shouldn't just drop one, and only populate one of the downstreams ? Or, should
> it replicate one input into both downstream topologies ? I can easily see
> interop issues here, and possibly even more basic transport issues from
> replicated data. 
> 

To use the functionality, the two streams of multicast must differ somehow
at the network layer.  For example, one uses (S1,G1) and the other uses
(S2,G2). Because the addresses are different, the forwarding trees will be
built using routes in the (multiple) topologies.  And once the forwarding
trees are built, the actual action of "forwarding" doesn't need to be
concerned with topologies.

> - if at any point the two multicast topologies go through a shared LAN or
> switch or any sort of multiple access network, how are the topological flows
> going to be kept distinct. (This is similar to the above, but the actual
> routers might be kept separate.)
> 
> - what happens with failovers ? If a link fails, SHOULD the backup links not
> collapse the topologies ? Again, I think some thought and text would be
> useful.
> 
> If there is a solution to this, it may be simple (you MUST use tunnels) or it
> may be complicated. In either case, some text here would be good.
> 

Marshall,  it would be good if you can suggest some text for us to use.
Thanks.


> Regards
> Marshall 
> _______________________________________________
> pim mailing list
> pim@ietf.org
> https://www.ietf.org/mailman/listinfo/pim


-- 
Yiqun


From ycai@cisco.com  Wed Aug 10 03:41:13 2011
Return-Path: <ycai@cisco.com>
X-Original-To: tsv-area@ietfa.amsl.com
Delivered-To: tsv-area@ietfa.amsl.com
Received: from localhost (localhost [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id B0DED21F86E0; Wed, 10 Aug 2011 03:41:13 -0700 (PDT)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -0.857
X-Spam-Level: 
X-Spam-Status: No, score=-0.857 tagged_above=-999 required=5 tests=[AWL=0.325,  BAYES_00=-2.599, HELO_EQ_FR=0.35, RCVD_IN_DNSWL_LOW=-1, RCVD_NUMERIC_HELO=2.067]
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 Tf13mgVXDd4L; Wed, 10 Aug 2011 03:41:12 -0700 (PDT)
Received: from p-mail1.rd.francetelecom.com (p-mail1.rd.francetelecom.com [195.101.245.15]) by ietfa.amsl.com (Postfix) with ESMTP id E887E21F86DE; Wed, 10 Aug 2011 03:41:11 -0700 (PDT)
Received: from p-mail1.rd.francetelecom.com (localhost.localdomain [127.0.0.1]) by localhost (Postfix) with SMTP id 9DC729B0009; Wed, 10 Aug 2011 12:42:33 +0200 (CEST)
Received: from ftrdsmtp1.rd.francetelecom.fr (unknown [10.192.128.46]) by p-mail1.rd.francetelecom.com (Postfix) with ESMTP id 90A439B0001; Wed, 10 Aug 2011 12:42:33 +0200 (CEST)
Received: from ftrdsmtp4.rd.francetelecom.fr ([10.192.128.49]) by ftrdsmtp1.rd.francetelecom.fr with Microsoft SMTPSVC(6.0.3790.4675);  Wed, 10 Aug 2011 12:02:40 +0200
Received: from mail pickup service by ftrdsmtp4.rd.francetelecom.fr with Microsoft SMTPSVC; Wed, 10 Aug 2011 02:17:45 +0200
Received: from omfeda06.si.francetelecom.fr ([10.98.3.82]) by ftrdsmtp4.rd.francetelecom.fr with Microsoft SMTPSVC(6.0.3790.4675);  Tue, 9 Aug 2011 22:50:25 +0200
Received: from omfeda14.si.francetelecom.fr (unknown [10.98.77.166]) by omfeda06.si.francetelecom.fr (ESMTP service) with ESMTP id 2D88CC803A; Tue,  9 Aug 2011 22:50:25 +0200 (CEST)
Received: from omfeda14.si.francetelecom.fr (localhost.localdomain [127.0.0.1]) by omfeda14.si.francetelecom.fr (ESMTP service) with SMTP id 1B9C92AC8B4; Tue,  9 Aug 2011 22:50:25 +0200 (CEST)
Received: from mail.ietf.org (mail.ietf.org [12.22.58.30]) by relais-inet.francetelecom.com (ESMTP service) with ESMTP id 7537D2AC8B2; Tue,  9 Aug 2011 22:50:24 +0200 (CEST)
Received: from ietfa.amsl.com (localhost [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id 0A0AB21F8B05; Tue,  9 Aug 2011 13:49:51 -0700 (PDT)
DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/simple; d=ietf.org; s=ietf1; t=1312922991; bh=jGrZ81X00klaf0McedkKG1XXA3WKvCKfLPXRji3SMD4=; h=Date:From:To:Message-ID:In-Reply-To:Mime-version:Cc:Subject: List-Id:List-Unsubscribe:List-Archive:List-Post:List-Help: List-Subscribe:Content-Type:Content-Transfer-Encoding:Sender; b=BQsNXm5WJ/Ajd8SK3xPLOlV6AX6bWHA0eoTSlCsbKkVYiMMR/1OlbAfb2lBjH/swG l7S6AsJs11s2QkLGXrhPdASwMp/E7A7MdWsQF5ffVSclQADB0iWdJcKaTCTobHPfIw XtwdoHAEl2QpkEslSHisooBLsdJR82kSjXvOAoIk=
X-Original-To: pim@ietfa.amsl.com
Delivered-To: pim@ietfa.amsl.com
Received: from localhost (localhost [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id 4CF6E21F8B04; Tue,  9 Aug 2011 13:49:49 -0700 (PDT)
X-Virus-Scanned: amavisd-new at amsl.com
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 BBtadf2gVD0A; Tue,  9 Aug 2011 13:49:48 -0700 (PDT)
Received: from rcdn-iport-8.cisco.com (rcdn-iport-8.cisco.com [173.37.86.79]) by ietfa.amsl.com (Postfix) with ESMTP id DA42821F8B02; Tue,  9 Aug 2011 13:49:47 -0700 (PDT)
DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/simple; d=cisco.com; i=ycai@cisco.com; l=10968; q=dns/txt; s=iport; t=1312923018; x=1314132618; h=date:subject:from:to:cc:message-id:in-reply-to: mime-version:content-transfer-encoding; bh=CKrLUywYoxwv2VZEj3FNiVTIxKhr3CKT4sw6BlHN+uY=; b=EOCmw3d7tH3DzQy/0WeQJwSSPGPruhOAqYcjZFGVPAEHhbNmN/+49JVt gPI3UWQfeCAqmyYmM9mVObEZjsYYn7g1D5D9I2w2/OA36lNEz+TBQQzvp FZ7sPwPW0QV/7W07Kukkx55hC/+Mx2tv4vgpVK89yvYr1GsTMxF4Jlvsl c=;
X-IronPort-Anti-Spam-Filtered: true
X-IronPort-Anti-Spam-Result: Ah8IABOdQU6rRDoJ/2dsb2JhbABCpz0Cd4FAAQEBAQIBAQEBDwEnAgExCwUNAQhnBjABAQQBDQUbB4dLBJ8QAZ5ahkYEh1yLKYUShF2DAIQc
X-IronPort-AV: E=Sophos;i="4.67,345,1309737600"; d="scan'208";a="11479663"
Received: from mtv-core-4.cisco.com ([171.68.58.9]) by rcdn-iport-8.cisco.com with ESMTP; 09 Aug 2011 20:50:16 +0000
Received: from xbh-sjc-221.amer.cisco.com (xbh-sjc-221.cisco.com [128.107.191.63]) by mtv-core-4.cisco.com (8.14.3/8.14.3) with ESMTP id p79KoGCj030223; Tue, 9 Aug 2011 20:50:16 GMT
Received: from xmb-sjc-216.amer.cisco.com ([171.70.151.184]) by xbh-sjc-221.amer.cisco.com with Microsoft SMTPSVC(6.0.3790.4675);  Tue, 9 Aug 2011 13:50:16 -0700
Received: from 128.107.161.230 ([128.107.161.230]) by xmb-sjc-216.amer.cisco.com ([171.70.151.184]) with Microsoft Exchange Server HTTP-DAV ; Tue,  9 Aug 2011 20:50:15 +0000
User-Agent: Microsoft-Entourage/12.26.0.100708
Date: Tue, 09 Aug 2011 13:50:21 -0700
From: Yiqun Cai <ycai@cisco.com>
To: Marshall Eubanks <marshall.eubanks@gmail.com>, IETF Discussion <ietf@ietf.org>
Message-ID: <CA66EB9D.AF52C%ycai@cisco.com>
Thread-Topic: [pim] tsv-dir review of draft-ietf-pim-mtid-08
Thread-Index: AcxW1fQVCxnlCmf+6kW+cOAFGDDVgQ==
In-Reply-To: <A9EFAA60-85DE-40F1-86C3-B9652E9CC0EA@gmail.com>
Mime-version: 1.0
X-OriginalArrivalTime: 09 Aug 2011 20:50:16.0307 (UTC) FILETIME=[F149CC30:01CC56D5]
Subject: Re: [pim] tsv-dir review of draft-ietf-pim-mtid-08
X-BeenThere: pim@ietf.org
X-Mailman-Version: 2.1.12
Precedence: list
Content-Type: text/plain; charset="us-ascii"
Content-Transfer-Encoding: 7bit
Sender: pim-bounces@ietf.org
Errors-To: pim-bounces@ietf.org
X-PMX-Version: 5.5.9.395186, Antispam-Engine: 2.7.2.376379, Antispam-Data: 2011.8.9.204214
X-Mailman-Approved-At: Wed, 10 Aug 2011 07:13:03 -0700
Cc: TSV Area <tsv-area@ietf.org>, pim@ietf.org, Transport Directorate <tsv-dir@ietf.org>
X-BeenThere: tsv-area@ietf.org
List-Id: IETF Transport and Services Area Mailing List <tsv-area.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/options/tsv-area>, <mailto:tsv-area-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/tsv-area>
List-Post: <mailto:tsv-area@ietf.org>
List-Help: <mailto:tsv-area-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/tsv-area>, <mailto:tsv-area-request@ietf.org?subject=subscribe>
X-List-Received-Date: Wed, 10 Aug 2011 10:41:13 -0000

Marshall,  thank you very much for the review.

Some of my comments are inline.


On 6/30/11 7:22 AM, "Marshall Eubanks" <marshall.eubanks@gmail.com> wrote:

> I've reviewed this document as part of the transport area directorate's
> ongoing effort to review key IETF documents. These comments were written
> primarily for the transport area directors, but are copied to the document's
> authors for their information and to allow them to address any issues raised.
> The authors should consider this review together with any other last-call
> comments they receive. Please always CC tsv-dir@ietf.org if you reply to or
> forward this review.
> 
> Note : I also reviewed the WG mailing list discussions for this document.
> 
> This draft is basically ready for publication, but has some issues that in my
> opinion should be fixed before publication.
> 
> There are, however, some issues that I think that should be addressed, either
> in a follow-on document, or a revision of this document.
> 
> draft-ietf-pim-mtid  introduces a new type of PIM Join Attribute [RFC5384],
> the MT-ID Join Attribute, that extends
> PIM signaling to cover multiple topologies and to enable the identification of
> which topology should be used when
> constructing a particular multicast distribution tree.
> 
> Multiple topologies have been used for some time in unicast (they are
> supported by IS-IS and OSPF), and also in multicast. The most common multicast
> requirement for multiple topologies is for video transport, where important or
> expensive video streams (say, of sporting events) are protected from
> disruption by multiple transport on completely different network paths. In
> that use, if a network link goes down or becomes congested or otherwise
> suboptimal, another source of the video data is already present and the
> display can be seamlessly switched to the other copy of the stream. This use
> case is alluded to in the document.
> 
> Another use case for multiple-topologies is to separate out different types of
> traffic (say, latency sensitive traffic from larger, bulk, flows). This is not
> alluded to in this document, and it is not clear to what extent this was
> intended. These multiple topologies may or may not result
> in the replication of data, and so may not be intended by the author. However,
> I would say that this could be used to support this, and so it will be. If
> there is some reason NOT to use this mechanism for this purpose, some
> description of the reasons why would be in order.
> 
> This document sets up a numerical variable, the  MT-ID Join Attribute, to
> assign to multicast topologies, to be used to separate different topologies
> for different paths. This number for a given multicast topology need not be
> the same as any unicast multiple topology identifiers, and similarly the
> multicast topologies and the underlying unicast topologies may be different.

You are right that there are indeed more use cases that can make use of the
functions described by this draft.  And some do exist in the field.

However, we (as co-authors) were very reluctant to list them.  We felt this
would be more appropriate to be described and published as whitepapers by
vendors, and we indeed have discussed them in public on behalf of our
employer.

It was our intention to try to keep this draft focused on the changes
required from a PIM protocol point of view.  I hope you will understand and
appreciate our viewpoints.


> 
> This document is sort, straight-forward, and relatively well-written. I had a
> few non-fatal issues.  (I would be glad to suggest some text for any of these
> issues to the authors if they interested.)
> 
> Minor Issues :
> 
> Section 3.2 
> 
> "- As shown earlier, this value is not required to be the same as the MT-ID
> used by the unicast routing protocols that contribute routes to the topology.
> In practice, when only one unicast routing protocol (such as OSPF or IS-IS) is
> used, the PIM MT-ID is RECOMMENDED to be assigned using the same value as the
> IGP topology identifier. Using the same example presented earlier, if
> every route in PIM 500 is contributed by OSPF 1000, it is RECOMMENDED to name
> this RPF topology as PIM 1000 instead of PIM 500. This is for the purpose of
> reducing management overhead and simplifying troubleshooting."
> 
> In the actual example, PIM 500 is not the same network topology as OSPF 1000
> (it has an extra link, to the source, obtained by MBGP). This is specifically
> mentioned previously in the text. So, this is NOT the same example as
> presented earlier and it is NOT clear that this recommendation applies here.

What we tried to say here is that,

1. If a PIM topology has routes contributed by different protocols, say OSPF
and MBGP,  it is better to use a different MT-ID for PIM.  For example, use
500 for PIM and 1000 for OSPF.

2. But if a PIM topology has all its routes contributed by a single
protocol,  then it is better to use the same MT-ID as the underlying unicast
protocol and in this case, 1000 is used for both PIM and OSPF.

Please let us know if the explanation helps and we will definitely
appreciate if you suggest some text here.

> 
> If this is left as is, it will certainly confuse some people. I would suggest
> adding text to clarify this, or creating another example where the
> unicast and multicast topologies are the same.
> 
> Section 4.2.3 
> 
> "- There is at most 1 PIM MT-ID attribute encoded. If there are multiple PIM
> MT-ID Join Attributes included, only the last one is accepted for this
> particular source. Processing of the rest of the Join message continues."
> 
> Is this a typo ? 
> 
> s/at most/at least/
> 
> would seem to be appropriate. (It says, "at most" 1, and then describes what
> do to if there is more than 1.) Otherwise, please explain what is meant.

We did mean there can be "at most" 1 MT-ID attribute.

But if there are more than 1 included, most likely due to an implementation
error, the rest of the paragraph suggests what an implementation should do.
Would it be better if we add "due to an implementation error"?

> 
> Section 6.
> 
> The Labels for Section 6.1 and 6.2 are the same. I suspect this is just a
> typo, and they should be
> 
> 6.1. PIM-Hello Options
> 
> 6.2. PIM Join Attribute Types
> 

Good catch.  Will fix.

> More Substantive Issues :
> 
> I had two substantive issues with this document.
> 
> 1.) The range for the topologies. It is highly likely that the users of this
> capability would want to use multiple topologies differently for different
> multicast address ranges. For example, use of two topologies is likely to be
> an expensive customer option, that not all customers would pay for, and so
> would not apply to all multicast addresses. Or, the backup typologies may
> (probably will be) tailored differently for different customers in different
> locations. Or, topologies might be tailored for different data types
> (different backups for real time video conferencing versus TV broadcast
> transmissions, say). This is very briefly alluded to in Section 3.1, but no
> consideration is given to issues arising from this.
> 
> For example, suppose I have two customers, and I wish to have one basic
> topology (say, PIM 100) for 239.10/16, and I wish to have two backup
> topologies for the two customers (say, PIM 1001 for 239.10.1/24 and PIM 1002
> for 239.10.2/24.) Can I have PIM 100 cover both the PIM 1001 and PIM 1002
> range ? Are there issues with this ? Would it be better to split PIM 100 into
> pieces to match 1001 and 1002 ? Nothing like this is discussed

The draft allows the above to happen.  Because as I indicated earlier, the
draft tries to focus on the protocol changes required to support the
functionality but not the deployment cases that can make use of it.

> 
> I can easily see interoperability issues arising from this (i.e., vendor A
> might allow PIM 100 to overlap PIM 1001 and PIM 1002, and vendor B might not).
> I think that some text to discuss the entire issue of ranging (presumably,
> that they can be allowed and overlaps SHOULD be supported) is needed.
> 
> Again, this could be out of scope here, but, then, where will it appear ?

I agree it is out of scope for the reasons I explained earlier but I don't
have a good answer what is the best practice to document all these cases.

> 
> 2.) There is very little discussion as to how to use this capability. Now,
> this might be considered out of scope for this document, but there will be
> issues, and I do not see a follow-on in the pipeline.
> 
> Issues include sharing of links and the general connections between
> topologies. It is likely that in real uses, a portion of the multiple
> topologies may be either physically shared or directly linked. It may not be
> possible to engineer out all single points of failure, for example an undersea
> cable. This is possible with the first use case (independent topologies for
> video) and likely with the second (different topologies for different types of
> content). 
> 

Yes, same as above.

> Issues I see here include
> 
> - the actual multicast data will not include any PIM attributes.  How, then,
> is a midpoint router receiving two copies of every packet know that it
> shouldn't just drop one, and only populate one of the downstreams ? Or, should
> it replicate one input into both downstream topologies ? I can easily see
> interop issues here, and possibly even more basic transport issues from
> replicated data. 
> 

To use the functionality, the two streams of multicast must differ somehow
at the network layer.  For example, one uses (S1,G1) and the other uses
(S2,G2). Because the addresses are different, the forwarding trees will be
built using routes in the (multiple) topologies.  And once the forwarding
trees are built, the actual action of "forwarding" doesn't need to be
concerned with topologies.

> - if at any point the two multicast topologies go through a shared LAN or
> switch or any sort of multiple access network, how are the topological flows
> going to be kept distinct. (This is similar to the above, but the actual
> routers might be kept separate.)
> 
> - what happens with failovers ? If a link fails, SHOULD the backup links not
> collapse the topologies ? Again, I think some thought and text would be
> useful.
> 
> If there is a solution to this, it may be simple (you MUST use tunnels) or it
> may be complicated. In either case, some text here would be good.
> 

Marshall,  it would be good if you can suggest some text for us to use.
Thanks.


> Regards
> Marshall 
> _______________________________________________
> pim mailing list
> pim@ietf.org
> https://www.ietf.org/mailman/listinfo/pim


-- 
Yiqun

_______________________________________________
pim mailing list
pim@ietf.org
https://www.ietf.org/mailman/listinfo/pim
