Sequence Count Value and SRR

Matt Wakeley matt_wakeley at agilent.com
Mon Jun 26 09:09:31 PDT 2000


* From the T10 Reflector (t10 at t10.org), posted by:
* Matt Wakeley <matt_wakeley at agilent.com>
*
There is even a better solution.

Change FC-FS and it's recovery qualifiers to include sequence_id as part of the
qualifier.  It's stupid to have to "retire" a sequence count range for all
values of sequence_id.  That way, error recovery could be performed using a new
sequence id and start from zero.

-Matt

"Zeitler, Carl" wrote:

> *
> * From the fc reflector, posted by:
> * "Zeitler, Carl" <Carl.Zeitler at compaq.com>
> *
>
> At the last meeting there was much discussion on Sequence Count and what
> values were used in retransmission.  For the current definition, a Sequence
> Count of 0 is required.  For future error recovery in Class 2, out-of-order,
> it would be desirable to use increasing Sequence Count to reduce error
> recovery time.
>
> I believe there is a simple solution.
>
> Define a bit (Word 3, Bit 9?) in the FCP service parameter page, PRLI
> request as follows.
>
> Give the bit a name: Sequence Count Usage (?)
>
> If the Sequence Count Usage Bit is 0, then the recipient of SRR shall
> retransmit Data Frames starting with a Sequence Count of 0.
>
> If the Sequence Count Usage Bit is 1, then the recipient of SRR shall
> retransmit Data Frames starting with the next consecutive value of the
> Sequence Count.
>
> With this added bit, current recovery can use continuously increasing
> sequence count and yet not have to remember the last used value.  If faster
> error recovery for out-of-order is desirable in Class 2 in the future, then
> this can be negotiated between Initiator and Target during PRLI.
>
> Regards, Carl
>
> Carl Zeitler
> Compaq Computer Corporation
> MS 150801, 20555 SH249, Houston, TX 77070
> Phone:281-518-5258 Fax: 281-514-5270
> E-Mail: Carl.Zeitler at compaq.com



*
* For T10 Reflector information, send a message with
* 'info t10' (no quotes) in the message body to majordomo at t10.org




More information about the T10 mailing list