The idea of a single link (consolidating I/O and storage) being a single point of failure doesn't stand up. You can very well have redundant 10GE uplinks just as much as today you can have redundant GE for I/O and redundant FC for storage. The point is to gradually remove FC from the picture.

This "unified fabric" idea makes a lot of sense, and Cisco was not the first to promote it. The main difference is that they are the only vendor that dares to claim it is ready now when it won't be before end of 2009. FCoE is not even close to be finalized. We'd need the "lossless ethernet" technology to be ready for this to happen. And today this is just a group of drafts with lots of open points to discuss. So others have been some time promoting it based on iSCSI for storage (which is ready and shipping) instead of FCoE.

I find it surprising to read in the press that Cisco people say they have a shipping FCoE product when the technology is not close to be ready. Maybe the Nexus line will have FCoE in the future but today it's just a big ethernet switch, and compared with others as just that, it lacks throughput, stability (new operating system) and features.

My 0.02...