Re: iSCSI: RE: Framing Discussion

[julian_satran@il.ibm.com]

Wayland,

Let's ignore for the first approximation anything that is not data.
Our assumption when outlining iSCSI was that all send data  (including most
of the iSCSI headers can be rebuilt from context).   The only context
information a sender has to keep is an association of a sender TCP sequence
number with either an iSCSI header or a SCSI buffer address and length.
The details will certainly vary with implementation.
In a loosely integrated solution the TCP stack will keep this context for
you.

In a tightly integrated solution you will have the TCP, on resend,
call-back the iSCSI layer to rebuild
headers and restate data addresses in which case iSCSI has to keep in its
tables some relation between the TCP sequence number and a specific data
piece (Task Tag, Data-packet-# etc.).

Regards,
Julo



Wayland Jeong <wayland@troikanetworks.com> on 26/12/2000 22:08:41

Please respond to Wayland Jeong <wayland@troikanetworks.com>

To:   Wayland Jeong <wayland@troikanetworks.com>, ips@ece.cmu.edu
cc:
Subject:  iSCSI: RE: Framing Discussion




I think I now understand the assumptions regarding the single-bit TCP
option
for indicating the presence of a PDU header in the current segment (thanks
Costa). It boils down to a hardware implementation that is tuned for the
best case (i.e. small amount of re-assembly memory on the NIC to park
fragmented PDU's with the assumption that the next aligned PDU is coming
shortly) and a method for dropping things into software when we are talking
to an ill-behaved NIC. Okay, I'll buy that.

Now, I have another dumb question. For direct data placement, the
discussions have been centered mostly around the need for alignment when
parsing PDU's on the receiving iSCSI TOE. One potential issue that has not
been discussed is the problem of how to handle re-transmission on the
sending iSCSI TOE.

From previous discussions, I am assuming that our goal is to avoid having a
network BWDP worth of memory on the NIC. The receiver can avoid this memory
by recovering PDU alignment in the TCP stream and using the self-describing
headers in the wire protocol (either iSCSI offsets or a RDMA shim layer) to
put the data directly in the buffer cache. On the sending side, we can DMA
directly from iSCSI descriptor CDB's into the TCP pipe using a hardware
path. But, unless we keep all of those un-acked TCP segment buffers around
in the NIC, it will be difficult to recover the context when we have to
re-transmit.

Let's suppose that we have an iSCSI TCP connection in which we have
multiple
outstanding I/O's. Thus, the byte stream has interleaved within it commands
and data from different I/O's. When we detect a dropped segment either
through normal TCP congestion or via SACK, how do we map the missing byte
block to the appropriate context? If we keep the segments around, then we
could match the missing segment easily and re-transmit. But that would
require the NIC to implement a BWDP's worth of transmit buffer memory.

To have the iSCSI TOE re-transmit directly from the buffer cache, it seems
that we would need some sort of context that would allow us to map a byte
window to a specific, meaningful point somewhere in the middle of a CDB
context. Essentially, you need enough context to be able to re-construct
the
TCP fifo since the memory in this fifo has since been effectively
re-allocated. Maybe this isn't too hard, but it sure sounds like a
difficult
problem for hardware to solve. But, as the software folks around here keep
telling me, "it's just gates" ;-)

-Wayland