re: FCoE is a great dead end

John Spiers — Fri, 11 Apr 2008 15:54:36 GMT

Mark, I’m glad you brought this up. I totally agree. FCOE is Fibre Channel’s only hope to keep pace with the Ethernet bandwidth progression from the current 10 Gig going to 100Gig. Of course the FC guys are counting their money from the potential of two lucrative upgrade cycles; have everyone replace their gear with 8 Gbps FC and then make them rip-and-replace it with FCoE gear. FCoE isn’t routable and doesn’t have congestion management, which relegates it to the four walls of the data center on pristine networks. They are trying to develop a bunch of hacks to Ethernet to provide congestion notification (the IEEE 802.1Qau), enhanced transmission selection (IEEE 802.1Qaz) and priority-based flow control (the IEEE PFC). What’s shocking about all this is that TCP/IP already has all this, plus it’s routable.

This is a classic example of vendors trying to re-design technology that already exists just to save their market. The only benefit I see, and it’s marginal at best, is the opportunity for lower cost connectivity to existing Fibre Channel gear, but why not just replace that gear with iSCSI and save yourself from having to purchase FCoE switches and FCoE HBAs for all your servers. With iSCSI you get all the promised benefits of FCoE with the additional benefit of a routable protocol, which gives you SAN based remote replication over standard networks without costly Fibre channel to IP bridge equipment. Not to mention a converged IP network.

re: FCoE is a great dead end

Nik Simpson — Thu, 10 Apr 2008 14:29:15 GMT

Couldn't agree more with you Marc. FCoE is all about preserving FC revenues and business model. I just love the way all the FC vendors are preaching from the same hymn book on this, reminds of that Ben Franklin quote " We must all hang together, or assuredly we shall all hang separately."

I particularly like the focus on preserving the FC management model, like that's anything worthy of preservation!