Forget FCoE - The War Is About Convergence
There has been a lot written about Fibre Channel over Ethernet (FCoE) the last few years but FCoE was merely an initial skirmish in the battle for the infrastructure. The major systems manufacturers are all placing their bets on everything over Ethernet and that is were the battle will be waged. 2011 may not be the year that you implement a converged infrastructure but it may be the year you decide which vendors you are going to use for convergence.
There has been a lot written about Fibre Channel over Ethernet (FCoE) the last few years but FCoE was merely an initial skirmish in the battle for the infrastructure. The major systems manufacturers are all placing their bets on everything over Ethernet and that is were the battle will be waged. 2011 may not be the year that you implement a converged infrastructure but it may be the year you decide which vendors you are going to use for convergence.FCoE, while it gets a lot of the hype, is really a subset of the conversation. It is the means by which Fibre channel infrastructures will move to Ethernet. Interestingly the core infrastructure suppliers are less concerned about which protocol you use than they used to be. We are seeing the same stance from the interface card vendors as well. Companies that a few years ago only made Fibre channel adapters are now shipping cards with no FC support, allowing you to add FCoE later. In both cases the plan is to make it easy for you to buy converged gear without having to converge right away. Start building a 10GbE infrastructure for networking, then add iSCSI and FCoE when and if you need to.
The ability to start with using converged infrastructure components as normal Ethernet networking devices first allows you to start building a converged infrastructure without actually doing any convergence. One process is to not do any converging until you have enough of these newer generation components that it makes sense to begin converging other protocols. This becomes a more logical migration strategy. It makes the move to convergence a slower process but may reduce some of the fear surrounding a more aggressive rip and replace strategy. The overall effect could be a wider acceptance of convergence.
The key for successful convergence on Ethernet is going to be fixing the limitations of the standard IP network and make it more data center ready. This means we are going to need to address the limitations of IP and Spanning Tree Protocol (STP). As we discussed in our recent article "Network Limitations To Cloud Computing and Convergence" data centers are becoming larger and server virtualization more widespread and we need to move to a more data center grade version of Ethernet, that is lossless and has a non-blocking architecture. The foundation for this will be based on TRILL but most of the major switch vendors will provide further enhancements. The HBA vendors will also be working to address limitations of Ethernet by offloading protocol processing and providing better virtual machine handling.
Convergence will happen, and there is every indication that is will be Ethernet based. The vendors are making it easy to get started by allowing you to buy products that can support multiple Ethernet protocols as you grow and upgrade your current Ethernet environment. The action plan for you now is to understand which infrastructure and card vendor is the best for your environment.
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.
About the Author
You May Also Like