There's an interesting article here by Swiss analyst George Crump on the merits of iSCSI vs. Fibre Channel over Ethernet (FCoE). Having written on this subject many times before, I wish Crump's e-mail server all the best over the next few days.
One of his key questions is that a chief benefit of Fibre Channel has always been that it gives the storage department control over its own network. Putting it on the Ethernet effectively hands that control over the networking group anyway, so why not go with lower-cost iSCSI? If the reason is that enterprises will want to leverage their existing Fibre Channel resources as long as possible, then at best, FCoE gives Fibre Channel a little breathing room until the bean counters start wondering why the organization is supporting two protocols instead of one.
My own take is that Fibre Channel probably will remain a going concern over the long haul because multi-protocol storage and network systems will lower the cost of maintaining both formats, and there will always be high-volume environments that simply need the more robust nature of Fibre Channel.
Crump also raises the question about maintaining quality of service on the network level to accommodate all those virtual machines that are headed your way. As I posted the other day, some in the vendor community are starting to wonder about that too.