2010 Year Of Fibre Channel-Over-Ethernet?

Will 2010 be the year of Fibre Channel-Over-Ethernet (FCoE)? I am always hesitant to predict that any particular year with be "the year" but I do think that FCoE will move out of conversation and testing phases and more into production.When it comes to something like infrastructure, changes come relatively slowly. The upgrade of previous storage infrastructures all came slowly and often did not occur until the new version of the infrastructure was the same price and the prior version. 2GB FC to 4GB FC as an example. There are three reasons why I think the switch to FCoE as a protocol of choice may come faster than previous infrastructure upgrades.

First FCoE like other infrastructure upgrades is a relatively seamless move, not much changes from a protocol perspective; the storage FC side is still managed the same as is the IP side. While FCoE does require new cards, cables and switches it does not require many new skills. Those cards, cables and switches are going to be purchased anyway as the environment expands. The decision point is will they be FCoE or the current protocol?

Second FCoE is ideally suited to be the infrastructure of choice for server virtualization projects where the limitations of a lack of a consolidated infrastructure are most telling. With the typical virtual host having two or four quad-port ethernet ports and typically two dual ported FC cards, the cable count in a rack of these servers quickly becomes a nest of unmanageable spaghetti. FCoE can reduce the size of that mess considerably.

Third there may be a more universal need for performance than we have seen in the past. Previous speed bumps in infrastructure were only needed by a few organizations that were pushing the envelope of performance. Other organizations simply got the performance boost as the new infrastructure became as inexpensive as the old. Now however, thanks again to server virtualization, data centers of all sizes are generating performance requirements that are saturating existing infrastructures.

Will 2010 be the year for FCoE? Certainly companies like Brocade, Cisco, Emulex and QLogic are all offering mature FCoE technology. However, as we discussed in our article "Planning for FCoE", most data centers are not going to do a wholesale change out to the protocol, it will be more of a gradual roll in. Many of these manufacturers are banking on 2011 to the "the year".

My opinion is that it will take five or more years for FCoE to become the dominant infrastructure in the data center but that each year we will see progress toward that domination. In truth however we will likely never be at a point where FCoE is the only infrastructure used. There will simply be too many point cases where other technology, like faster dedicated IP, faster dedicate FC or even I/O virtualization, may be more appropriate.

IT managers would be well advised to begin production deployments of FCoE in the coming year but to do so gradually as the need warrants. With a gradual roll out of the technology however it is important to make sure the vendors that supply your FCoE infrastructure continue to support the dedicated versions of the protocols. Finally keep your eyes open on I/O Virtualization which may be a cost effective alternative.

Track us on Twitter: http://twitter.com/storageswiss

Subscribe to our RSS feed.

George Crump is lead analyst of Storage Switzerland, an IT analyst firm focused on the storage and virtualization segments. Find Storage Switzerland's disclosure statement here.