Let’s face it, Fibre Channel over Ethernet (FCoE) has not been the roaring success that vendors hoped it would be. Why? Probably the number one reason is that it was too expensive. Users expected FCoE to be cheaper, much cheaper, than regular fibre channel and were somewhat put off when it was not. Although vendors came up with elaborate return on investment (ROI) models to justify the extra expense, in reality the upfront costs were prohibitive. As a result FCoE, while being adopted, is being adopted at a much slower pace than originally expected.
Part of the challenge was that moving to FCoE required new switching fabrics that were only available from a few vendors and there was concern about interoperability between those switches. In fairness, these switches did offer some interesting features to extend the usefulness of layer-two networks, like TRILL, but the big challenge with requiring a new switching fabric is its cost. Sure it could run alongside an existing fabric but these switches tended to sell at a premium vs. more traditional fibre channel and certainly more than IP switches that the iSCSI and NFS protocols would use.
The other part of the challenge was that FCoE required special Converged Network Adaptors (CNAs) that also were priced at a premium vs. traditional FC adaptors or standard off the shelf 10GbE cards. CNAs, like FCoE switches, offered extended functionality, such as the ability to support a mix of standard 10GbE, iSCSI (with an offload capability) and FCoE itself. Unlike FCoE switches the cost of CNAs have come down significantly since their introduction. This has been driven in large part by the competitive nature of the adapter card market and Intel’s introduction of a software driver that could allow any 10GbE card to communicate via FCoE.
FCoE targets were less of an issue. Storage manufacturers, as long as they had 10GbE ports on their storage systems, could easily adopt FCoE if there was demand.
The net of this discussion is that the bottleneck to the adoption of FCoE seems to be the advanced switching requirement. This is what Jeda Networks is hoping to address with their Software Defined Storage Network solution (SDSN).
At a simple level Jeda’s product allows virtually any switch, once qualified, to act as an FCoE switch. That means much less expensive standard 10GbE switches could be part of your FCoE plan going forward. With a Jeda Networks switch in place suddenly FCoE goes from the most expensive storage networking design requiring a fancy ROI tool to justify, to the least expensive storage network on the market. And it does this while delivering rock-solid fibre channel reliability and performance.
Like other software defined solutions, Jeda Networks has essentially abstracted the switch intelligence from the switch hardware, which means that any switch and a mixture of switches could form your storage network backbone. For the storage network designer this brings massive cost control capabilities.
Storage Swiss Take
There are a lot of storage protocols to choose from. We are seeing a resurgence in fibre channel, thanks to the Gen 5 initiative, the adoption of iSCSI remains strong and NFS, especially in virtual environments, has its strengths as well. But there is clearly a need for something in between that approaches the cost effectiveness of iSCSI/NFS but has the capabilities of fibre. FCoE was supposed to be that something, but high pricing dampened initial enthusiasm. FCoE needs a bit of a reboot from a marketing perspective and Jeda Networks may be the company pressing Ctrl-Alt-Del.
Jeda Networks is not a client of Storage Switzerland