Date: Sat, 9 Aug 2003 13:43:55 -0500
Subject: Re: iSCSI/iWARP drafts and flow control
Content-Type: text/plain; charset=US-ASCII; format=flowed
Mime-Version: 1.0 (Apple Message framework v552)
Cc: <ips@ece.cmu.edu>, <rddp@ietf.org>
To: Mike Ko <mako@almaden.ibm.com>
From: Caitlin Bestler <cait@asomi.com>
In-Reply-To: <OF71F9EAE1.1E9FCF97-ON87256D7C.00755A93@us.ibm.com>
Message-Id: <6E2BBF37-CA99-11D7-9E17-003065D48EE0@asomi.com>
Content-Transfer-Encoding: 7bit
Sender: owner-ips@ece.cmu.edu
Precedence: bulk


On Friday, August 8, 2003, at 04:32 PM, Mike Ko wrote:

> Since the Asynchronous Message is associated with a StatSN, instead of
> waiting for the initiator to send "a command with CmdSN that was only
> enabled after the non-responding target message", another possibility 
> is
> to wait for the initiator to send any PDU containing the ExtStatSN 
> field
> that acknowledges that the particular StatSN has been received.

Even better.

I am not favoring any specific mechanism, just saying that
mechanisms exist -- and that they do not have to be an
explicit "protocol".

Note that none of the algorithms described, including your
suggestion, require the receiver to calculate or issue
credits. They merely have to replenish buffers in manner
consistent with the rules. They can also be sloppy about
it if they have slop in their buffers.

The rules suggested shift the burden of flow control
to the sender, where it should be. Rule enforcement
at the sending side can also be quite lax if other
implementation specific rules limit the number of
untagged messages that could be sent.





Caitlin Bestler - cait@asomi.com - http://asomi.com/