Will FCoE Be 2009's Next Victim?

The bad economy this year has laid waste to many IT budgets and, with them, many well intentioned infrastructure upgrades. Big projects that will make life better or provide a long-term ROI are being set aside in favor of extending what is already there and driving a faster ROI.

Fibre Channel over Ethernet (FCoE) may be a good example of something that is set aside for now. As a result, the battle between FCoE vs. 8GB Fibre vs. iSCSI vs. [fill in the blank] may be rendered moot for the next few years.

Interestingly, there is a need for improved performance in many Fibre Channel infrastructures right now. Some of this is being driven by the neverending need for increased performance, and some is being driven by the increased implementation of solid-state disks (a 30x faster drive needs faster pipes). But much of the performance demand is being driven by server virtualization.

Server virtualization is producing an increase in virtual machines per physical host. The increase is driven in part by growing comfort and confidence in the technologies, but it is also being driven, ironically, by the down economy. The risk of stacking yet another virtual machine onto a physical machine is one worth taking if it can save on precious budget dollars.

The problem is, this puts additional load on all the components of that physical server, with one of those being the storage I/O path. Also, these workloads couldn't be more random. Unlike in the past, where it was one application per HBA, now you can have 20, 30, or 40 applications contending for storage I/O bandwidth. All this drives the need for increased I/O performance, and we are due for some quality-of-service capabilities from our storage I/O infrastructure (more on that in an upcoming entry).Returning back to the original point, we have decreased budgets that are going to make dramatic infrastructure changes undesirable. Yet we have more legitimate reasons than ever for increasing storage I/O performance. Now what?

The logical path for many data centers will be to continue to invest in the two physically discrete networks (IP and FC). For storage, that means an upgrade to 8GB Fibre Channel for right now. Look for end-to-end QOS types of capabilities to help enhance server virtualization projects, and then look for technology that provides an incremental and cost-effective transition path to FCoE, when and if it makes sense for your data center.

At some point we are going to have to explore whether there is really a cost savings to the consolidation of the physical layers within the data center. But 2009 is not that year.

Please view our seminar on Primary Storage Optimization.

George Crump is founder of Storage Switzerland , which provides strategic consulting and analysis to storage users, suppliers, and integrators. Prior to Storage Switzerland, he was CTO at one of the nation's largest integrators.6668