RE: small change in the connected mode draft

I think that this is reasonable amendment. Being practical, I don't
think that IPoIB-CM will be widely used over WANs or long latency
fabrics. To me it looks something very data center oriented, something
like Jumbo frames in Ethernet.=20
 =20
Other than that, IB is mostly a lossless fabric, so I wouldn't expect to
see that many of retransmissions in RC anyway.
 =20
Dror=20

> -----Original Message-----
> From: H.K. Jerry Chu [mailto:[email protected]]=20
> Sent: Wednesday, March 22, 2006 2:05 AM
> To: [email protected]
> Cc: [email protected]; [email protected]; [email protected]
> Subject: [Ipoverib] small change in the connected mode draft
>=20
>=20
> Hi folks,
>=20
> Allison Mankin, the transport area AD who brought up some=20
> concern during IESG review of the connected draft regarding=20
> simultaneous retransmissions at different layers, has=20
> suggested and Vivek agreed to the following change to section=20
> 7.1 "A Cautionary Note on IPoIB-RC".
>=20
> The revised section reads like this:
>=20
> 	The RC mode of InfiniBand guarantees in-order delivery of
> 	packets. Every message transmitted over the RC connection is
> 	broken into physical MTU sized packets by the RC connection. If
> 	any packet is lost, it is retransmitted until the complete
> 	message is exchanged. Therefore, there is a possibility of an
> 	upper transport layer experiencing a timeout, while the RC layer
> 	is still in the process of transferring the complete message.
> 	TCP will view the timeout as an indicator of congestion and
> 	enter slow-start thereby affecting throughput drastically
> 	[RFC2581]. Other upper layer protocols might insert
> 	retransmissions into the fabric adding to the already existing
> 	congestion.
>=20
> 	The applicability of Infiniband reliability is on a fabric
> 	with short latencies (not wide area).  Therefore, the RC timer
> 	values should be short compared with the starting minimum
> 	time values used by the upper end-to-end transports.  In
> 	addition, because the RC mode does not have measurement
> 	based reliable transmission, its use over fabrics with long
> 	latency or very dynamic latency may be a concern for congestion-
> 	aware traffic traversing those fabrics.
>=20
> If you have any comment/issue on the proposed change please=20
> post them to the list before COB this Friday (3/24/06). I'd=20
> like to move the draft past IESG review so we can wrap up the WG asap.
>=20
> BTW, please be informed that my email address will change=20
> after this Friday. My new address is [email protected].
>=20
> Thanks,
>=20
> Jerry
>=20
> IPoIB co-chair
> Sr. Staff Engineer
> Solaris Core Networking
> Sun Microsystems, Inc.
> U.S.A
> (650) 786-514
>=20
>=20
>=20
> _______________________________________________
> IPoverIB mailing list
> [email protected] https://www1.ietf.org/mailman/listinfo/ipoverib
>=20
>=20
>=20