Re: ISCSI: Why the use of the Urgent Pointer

[Matt Wakeley <matt_wakeley@agilent.com>]

Dick Gahan wrote:

> >There could be an Urgent Pointer use negotiation mechanism, where if an iSCSI
> >implementation indicates during login that it wants (requires) to receive the
> >urgent pointer, the other side must send it, even if it does not want to receive
> it.
>
> I wouldn?t call this negotiation. It sounds more like coercion, the other side
> does not have an option.
> The function MUST be provided when requested thus not an option.

All the comments have been related to not wanting to receive and process urgent
data.  So, with this option, if an implementation doesn't want to receive it, fine,
just tell the other node it's not wanted.  But if that other node wants (needs) to
receive it, send the iSCSI PDUs using the proposed urgent pointer.

>
>
> You also state that:-
>
> >In order for iSCSI to succeed it must be able to be cost competitive with
> >other solutions (10Gig Fibre Channel, etc).  This means that it must be
> >implementable without putting gobs of (costly) memory on the adapter cards.
> >In order to avoid this memory requirement in the presence of lost frames, the
> >adapter will either have to place data in host memory were it belongs
> >*requiring a framing mechanism* or throw it away.  Fibre Channel adapters
> >today are able to operate in this manner today without a lot of memory because
> >they can place data that arrives out of order directly where it belongs in
> >host memory [because framing is provided].
>
> This TOE adapter you mention seems dedicated for iSCSI. Other applications
> that talk gig Ethernet over TCP do not seem to use this TOE. If they did use it
> what does the TOE do for those other (not iSCSI) streams when data is lost ?.
> They have not an URG framing mechanism built in and the adapter has not much
> memory to buffer up data while the missing piece is retransmitted.
> Are you advocating a second gigabit NIC with a TOE function which has memory ?
> or a general purpose gigabit adapter which uses host memory and slow host tcp/ip
> stack ?.

> So I don?t see any reason to support your proposal right now.

It sounds to me that you think my "TOE" requires the urgent pointer.  My "TOE",
just like all other "TOE"s, does not require framing.

It is the "iSCSI" engine sitting on top of the "TOE" that requires the framing.
This is because in the storage world, received data is placed into pre-allocated
buffers that were supplied specifically for the I/O. iSCSI needs the framing in
order to place the data where it belongs when it comes off the link, thus avoiding
buffering and copying.

-Matt Wakeley
Agilent Technologies

>
>
> Dick Gahan
> 3Com