Cisco's Unified Computing announcements last week provided a lot of ammo for friends and foes. On the up side, it's hard to argue with the message that existing server, storage and networking architectures render a virtualized data center hard to manage.Certainly the need to attach current systems to as many as three distinct fabrics (LAN, SAN and a high performance network like InfiniBand), all with inherent limitations and with addressing schemes never imagined for highly dynamic, virtualized environments is a huge problem.

However Cisco's discussions of an open-standards-driven environment for both a unified network and a unified compute environment sounds suspiciously like its early proclamations for its NAC alliance. That alliance, which is driven by Cisco proprietary protocols, went from "all players welcome" to a limited membership when Cisco tossed ConSentry for making products that looked too much like a competitive switch.

Another problem afflicting the original notion of NAC could also be a problem here, and that was the eye popping price tag for a full blown Cisco NAC framework implementation. Price was such a deal breaker that Cisco had to eventually roll out a NAC appliance that was price competitive with other products.

The lesson here is that the devil is indeed in the details and Cisco has a mixed track record in reading what the market will bear both in terms of the cost of new technologies and in terms of vendor lock-in. But where NAC competition was made up of largely small players - a group with whom Cisco rarely plays well, the competitors in the data center have at least as much weight to throw around as does Cisco.

HP, with its sizeable data center influence, has been steadily moving its ProCurve switching unit closer to its server and storage divisions, and IBM and Sun both have a vested interest in choice in the data center. VMware and Microsoft each have their own vision of the virtualized data center, and Juniper and Brocade though smaller, have both made unified data center architectures announcements of their own.

All this bodes well for competition in the brave new world of unified data center architectures, as long as the competition is within standards, and not with standards. Cisco in particular needs to reign in its tendency to create ad hoc proprietary standards. It's a false argument to say that standards bodies are too slow to meet your needs, particularly when you've been claiming that your current vision is the result of years of research.

For enterprise architects, it's likely the unified data center architectures will take a path something like VoIP adoption. While the benefits of a single fabric are unmistakable, rip and replace isn't needed; you can adopt a fabric through the normal cycle of equipment amortization and replacement. Particularly in terms of standards, management and security, there's so much to be worked out and learned, that your amortization cycle should be viewed a friend, and not something to be worked around.