Quite the Tempest in a Teapot. It reminds me of the choice between chocolate and peanut butter, a false choice in my opinion so I stock both because sometimes I want one, sometimes the other and sometimes I want both.
Rather than contribute to the Teapot Tempest, I thought I’d talk a bit more about customer choice when it comes to storage networking, converged networks and data center design. I’ll finish up by providing some examples of Brocade solutions customers are choosing today since I think focusing on what our customers are buying is an effective way to address the conjectures and opinions put forth by J and Steve.
Here are some questions storage and network architects should ask themselves when it comes time to expand an existing storage network or design a new one.
If you use iSCSI, does it do everything you need?
If you use Fibre Channel, is there anything you need to do that Fibre Channel can’t do but FCoE can?
Compared to the incumbent technology, when does a new technology become a compelling choice?
The first question acknowledges the fact the architecture and technical knowhow for iSCSI is different than FCoE and Fibre Channel. If you have invested in iSCSI and it works for your next project, you’re done. With 10 GE interfaces and lossless Ethernet available (and even better, Brocade VCS Ethernet fabrics for best in class iSCSI networking), you’re good to go, so get going.
To the second question, I think many Fibre Channel administrators would answer “not much” when it comes to speeds, feeds, robustness, security and technology standards. But the cable mess and associated cost in server racks that is created by the growth of x86 server virtualization can be a problem. FCoE can address that problem by converging traffic on a single 10 GE link so it’s logical to consider it. But in the end, FCoE is Fibre Channel with a straight forward substitution of Ethernet as the link layer. Both require you to understand Fibre Channel and to know how to administer it. If you have trained Fibre Channel networking staff, no problem, pick which works best for the project goals and your overall data center strategy. Different applications have different needs so choice is a good thing.
I firmly believe that thoughtful technology choices are grounded in considerations of economics, risk, skills and time. When a new technology arrives, it will have a place in a production data center when it offers compelling value across those four dimensions.
Business goals drive IT projects and IT projects drive technology adoption. FCoE technology adoption will happen the same way, over time as projects create the opportunity. I think a driver for FCoE is the growth of server virtualization and the continuing cost reduction 10 GE server connectivty. For many customers, getting control of the cabling chaos in server racks when scaling up virtualization is when FCoE gets considered. For others, iSCSI is the better choice, particularly with 10 GE interfaces available in the storage arrays. Using a VCS Ethernet fabric for lossless, low latency, simple to configure access and aggregation layers is a great way to build a converged iSCSI network. And, based on industry analyst tabulations of the type of shared storage deployed in the majority of data centers, Fibre Channel storage is still the primary choice, so that’s what customers continue to use to store their virtual machine files and application data.
I believe customers make a choice about iSCSI, FCoE, Fibre Channel or NAS based on an evaluation of THEIR economics, risk, skills and time. That’s why many data centers have more than one storage network deployed, because it made sense to them at that time for that particular project.
Now the reason J wrote his blog is that he was upset with Steve for not truthfully representing the facts about FCoE. Blogging is about the free flow of opinions, perspectives, prejudices and chosen facts to support the case at hand, so blogs beget blogs, and that’s part of the entertainment value. Fair enough.
However, in his response, J references a blog he wrote back in November 2010 about the meaning of end-to-End FCoE which contains a critique of Brocade’s claims of delivering the first End-to-End FCoE solution when we announced the availability of VCS Ethernet fabrics. But, it isn’t November 2010 anymore, and this out-dated blog by J no longer accurate what Brocade offers.
Since old blogs never die, I wanted to set the record straight based on Brocade’s offerings today. Here is a short list of the “end-to-end” (from server to storage including Brocade adaptors) data center solutions you can buy from Brocade.
FCoE in the server connecting to FCoE AND Fibre Channel storage by internetworking our VDX 6730 to our Fibre Channel family products using Fibre Channel routing.
iSCSI or NAS storage (GE and 10 GE) with lossless VCS Ethernet fabrics using the VDX 6710 and VDX 6720.
Single server adaptor for wire once servers providing 10 GE or Fibre Channel connectivity using AnyIO technology from Brocade with the Brocade 1860 Fabric Adaptor.
Integrated management of IP, Fibre Channel and converged networks with Brocade’s Network Advisor including automatic discovery of virtual machines and automatic configuration of network policies in a VCS Ethernet fabric.
Brocade’s data center portfolio delivers a full range of technology options so our customers don’t have to waste time on “Tempest in a Teapot” arguments over false choices between FCoE OR Fibre Channel. When you choose Brocade, you can use either with confidence since both are based on our proven Fibre Channel technology or you can go with iSCSI and continue to take advantage of VCS Ethernet fabrics.
Whatever storage technology makes sense for you, Brocade provides a rainbow of options … even double rainbows, it that’s what you want.