Date: Mon, 18 Aug 2003 10:34:45 -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: wrstuden@wasabisystems.com
From: Caitlin Bestler <cait@asomi.com>
In-Reply-To: <Pine.NEB.4.53.0308162206260.77@neverwinter.home-net.icnt.net>
Message-Id: <7E714293-D191-11D7-B04A-003065D48EE0@asomi.com>
Content-Transfer-Encoding: 7bit
Sender: owner-ips@ece.cmu.edu
Precedence: bulk


On Monday, August 18, 2003, at 07:24 AM, wrstuden@wasabisystems.com 
wrote:

> I'm not saying we can't just pick a number. But if I understand things,
> one advantage of per-session negotiation might be that a target with 
> fixed
> buffering could offer different command windows to initiators that 
> wanted
> different numbers of immediate commands available. For an initiator 
> that
> is more comfortable advancing CmdSN (is happier with a lower number of
> credits per CmdSN), a target could divy up its buffers to allow a wider
> command window.

The cost of per-session negotiation is minor, so I suspect that
almost any solid example of an application or scenario that
would benefit from a non-standard number of credits would
be convincing.

But given that the flow control is imposed on the sending side,
and that these are rare packets, it isn't that easy to come up
with such a scenario.

I'm not opposed to the idea, but if a parameter is going to be
listed as negotiable in the draft there should be at least a
paragraph hinting at why you might want to vary it (even if
it was only on the mailing list and doesn't make it into
the draft).