There's no doubt that the server market needed a shake-up. Incumbent players were still focusing on including hard drives and lots of networking ports on their servers, when they should have been packing in as much processing and memory as possible. Cisco's virtualizing the BIOS is a nifty move, and it would have been even better if co-developer Intel had pushed it as an industry-wide feature so that management tools would have universally understood it. The same can be said for the custom silicon that allows Cisco to pack 384 GB of memory where most manufacturers put 192 GB or less.
As a concept, these features, along with the consolidated networking at the heart of Cisco's UCS, is exactly what future data centers should look like. But the devil is in the details, and the reality is that Cisco has overestimated the willingness of customers to accept the vendor lock-in that comes with implementing a system like UCS. Giving up existing management tools and methodologies is a very hard sell, particularly when at least some of the technical advantages can be had through other means.
Fundamentally, Cisco is offering a method of abstracting hardware so that operating systems and applications can be easily moved around the data center as needs dictate. It sounds a lot like virtualization, and if you can do almost everything with your VMware investment, why would you take on another disruptive technology? For almost everyone, the answer is that you wouldn't.
That's the bet that Dell, HP, and IBM are making as they continue to make new product announcements. In particular, Cisco is eyeing HP, which brought its ProCurve switching products into its server and storage group. If Cisco had stuck with its avant-garde view of how things ought to be in the data center, then I'd say HP would have had a field day satisfying the 95% of the market that simply doesn't have avant-garde aspirations.