Re: iscsi : DataPDULength can differ in each direction.

[Dave Sheehy <dbs@acropora.rose.agilent.com>]

Santosh,

> Is this a feature that you believe is going to be useful ? 

Yes, I do. I already described the specific scenario in a prior email.

> I proposed
> this because I believed the way it is currently done is incorrect. The
> DataPDULength negotiation should ideally follow the model of the FC
> PLOGI common service params "receive buffer length" which an initiator &
> target exchange. In essence, each side tells the other what is the
> maximum length it can receive.

I agree. In retrospect it seems obvious.

> However, I'm also pragmatic and realize that several vendors have
> already gotten their implementations done and so, I did'nt push for this
> strongly. Do you believe this is something that is going to be useful
> and implementations will require specific tweaks to the PDU lengths they
> can use ?

Since it is a change in behavior it will require tweaks as you say but not
significant changes. I believe that the gains make it worthwhile.

Dave