Ever since FCoE was first announced with much fanfare a few years ago at the spring Storage Networking World 2008, it has always struck me as a technology looking for a problem to solve. Now nearly three years later it seems that the largest advocate for FCoE (arguably Cisco) may have finally found a practical use case for it. But the question that has since crossed my mind is, “Does anyone who has implemented FCoE in their data center really know that FCoE is the protocol that Cisco is using under the covers in support of its SAN infrastructure?“
The insight I got into this practical use case for FCoE came as a result of a conversation I had with a long time friend of mine who I had not seen in nearly five years but ran into this past Monday at the regional Kansas City VMUG even. He and I used to communicate quite a bit when I still worked for First Data and he still worked for Brocade. But after he left Brocade to work for Cisco and I left First Data to start DCIG, we had not been in contact until this past Monday.
So as we were catching up as to what each of us had been up, he started to explain how he is selling Cisco’s Unified Computing System (UCS) products such as its UCS B-Series Blade Servers, its UCS 5100 Series Blade Server Chassis, and its UCS 6100 Series Fabric Interconnects.
He said that to date he has had good success selling this solution into customer accounts for a number of key reasons. The advent of server virtualization has made server hardware essentially a commodity so many of the accounts into which he is selling do not really care whose server hardware their OS is running on. It is this attitude that in part inspired Cisco to build its own server hardware platform. The advantages that server platforms from Dell, HP, IBM and others offered were not sufficient in the minds of many of his customers to give them enough reasons to stay with them as opposed to switching to Cisco.
However just offering a server hardware platform that cost the same or even a little less likely would not alone prompt them to switch to Cisco. Instead what has been winning him a lot of deals was the much lower price of the Cisco UCS 5100 Series Blade Server Chassis as compared to its competitors.
He has routinely found that the Cisco Series Blade Server Chassis comes in at 10 – 20% of the price of its competitors which, if all other things in a competitive situation are equal, is enough for Cisco to get the nod over its competitors.
It was also at this point that he shared that the protocol that Cisco is using to provide SAN connectivity for its blade chassis is FCoE. This factor may, in part, account for the significant discount that Cisco can offer to its customers versus its competitors. While he did not share if that was the case (nor I did not think to ask about it when we talking,) in thinking about it not having to install extra FC HBAs or provide for FC connectivity may account for Cisco’s ability to offer its blade server solution at a much lower price.
This might also explain why Cisco has been so cordial towards NetApp and is seen on almost even slide from NetApp that contains NetApp’s recently announced FlexPod for VMware architecture. All NetApp products support the FCoE protocol which eliminates the need for organizations to spend extra money on a FC switch or director. This is relative as the Cisco chassis can now communicate directly with the NetApp FAS Series products over Ethernet.
So does this mean that FCoE suddenly has a bright future ahead of it? That is unclear. Someone shared with me that to add the FCoE protocol to most Ethernet switches still requires that users purchase a separate license to turn on this functionality. On a more positive note, another storage architect said that he has deployed it and has had good success with using it. But as far as I know, only a few (as in very few) storage vendors support the FCoE protocol (NetApp has supported FCoE for some time while Compellent just announced support for it last week on its Storage Center 5.4 release.)
After three years it appears that FCoE enthusiasts finally have a practical use case that they can point to as a reason to deploy FCoE. But that said, it is not clear whether or not those organizations who are deploying FCoE really understand (or care) that it is FCoE being implemented underneath the covers. It may just be that they are focusing on the lower price point for their blade servers and then only finding out after the fact that FCoE has been implemented in their infrastructure.