Sean Hefty wrote:
IMO, the fact that TCP implements reliability doesn't mean it's unnecessary in underlying layers. For example, wireless typically adds reliability at the link layer because the link itself is so unreliable. If adding in reliability in the underlying layers improves overall performance, then it makes sense to add it, independent of the upper level protocol.
For applications such as UDP based voice/video streaming, retransmitting this frame over and over might cause degradation in the service. With the current implementation maybe the 1st problem is the RC nature and the retries are only the 2nd but why add them.
For for link (and below) layers reliability, IB has this 8-to-10 bits encoding, CRCs, L2 credit management, etc. So much is done below the IB transport legs anyway...
Or _______________________________________________ general mailing list [email protected] http://lists.openfabrics.org/cgi-bin/mailman/listinfo/general To unsubscribe, please visit http://openib.org/mailman/listinfo/openib-general
