RE: iSCSI minimum PDU length

["Julian Satran" <Julian_Satran@il.ibm.com>]

Shailesh,

With a high level of enthusiasm  the consensus was on 512 (2 voices for 
512 against 1 voice for 1024).
I am not sure about what you are asking in the second part of the message.
The 512 refers to the data-segment of an iSCSI PDU and states the minimum 
that MUST be supported.
If you are asking about TCP it says nothing about TCP (TCP is a stream of 
bytes). In practical terms you would like to have TCP support packets that 
are at least 560 (or 564 if CRCs are enabled - 512+48).

Julo




"Shailesh Manjrekar" <shaileshm@aarohicommunications.com>
16-11-01 01:44
Please respond to "Shailesh Manjrekar"

 
        To:     John Hufferd/San Jose/IBM@IBMUS, Julian Satran/Haifa/IBM@IBMIL
        cc:     <ips@ece.cmu.edu>
        Subject:        RE: iSCSI minimum PDU length

 

Julian,

Have we reached a consensus whether the minimum PDU size would be 512 or
1024. Also for command PDU's ( 48bytes + header ) does it mean that we
would end up using a MTU size ( 1500 bytes ) TCP segment for this. This
would add segment processing overheads at offload engine for just
48bytes + header.
Please clarify,

Regards,
Shailesh.
Aarohi Communications.
-----Original Message-----
From: owner-ips@ece.cmu.edu [mailto:owner-ips@ece.cmu.edu] On Behalf Of
John Hufferd
Sent: Thursday, October 25, 2001 2:36 AM
To: Julian Satran
Cc: ips@ece.cmu.edu
Subject: Re: iSCSI minimum PDU length

If you are only talking about Data PDUs, then at least I understand.

.
.
.
John L. Hufferd
Senior Technical Staff Member (STSM)
IBM/SSG San Jose Ca
Main Office (408) 256-0403, Tie: 276-0403,  eFax: (408) 904-4688
Home Office (408) 997-6136
Internet address: hufferd@us.ibm.com
---------------------- Forwarded by John Hufferd/San Jose/IBM on
10/25/2001
02:34 AM ---------------------------

John Hufferd/San Jose/IBM@IBMUS@ece.cmu.edu on 10/25/2001 02:12:07 AM

Sent by:  owner-ips@ece.cmu.edu


To:   Julian Satran/Haifa/IBM@IBMIL
cc:   ips@ece.cmu.edu
Subject:  Re: iSCSI minimum PDU length




Julian,
Perhaps I do not understand what you are getting at here, but since
there
will be implementations with only R2T type Sessions, then there should
be a
lot of PDUs that just carry the Command (48 bytes + digest) it seems
that a
1024 is kind of big.  What am I missing?
.
.
.
John L. Hufferd
Senior Technical Staff Member (STSM)
IBM/SSG San Jose Ca
Main Office (408) 256-0403, Tie: 276-0403,  eFax: (408) 904-4688
Home Office (408) 997-6136
Internet address: hufferd@us.ibm.com


Julian Satran/Haifa/IBM@IBMIL@ece.cmu.edu on 10/24/2001 10:34:38 AM

Sent by:  owner-ips@ece.cmu.edu


To:   ips@ece.cmu.edu
cc:
Subject:  iSCSI minimum PDU length



With no votes against we have settled (again) on single PDU length (per
connection, per direction) for all types of PDUs.
But I think we erred on the low side by suggesting 64 as a minimum. It
is
low (it was mentioned) and bad for text request/response.

How about settling for 1024?

Julo