At many data centers, the upgrade from 2 Gbps to 4 Gbps happened almost by default. The price of the 4 Gbps components had dropped to the price of the 2 Gbps components. Many IT organizations really didn't need 4 Gbps Fibre Channel, they just got it because, for the most part, that was what was available.
But 8 Gbps Fibre Channel is different. When 4 Gbps arrived, there wasn't performance-hungry hardware placing huge demands on storage I/O. But solid-state disk (SSD) has changed that dynamic. It can easily saturate multiple 4 Gbps connections today and will take full advantage of the bandwidth that 8 Gbps can provide. But faster-responding disks make up half of the justification for a faster protocol; servers that are making more and faster requests of those disks are the other half, and that's where server virtualization comes in. It stacks up multiple servers (in the virtual sense) on a single physical server. All of those virtual servers could be making simultaneous requests to the storage infrastructure. The faster that infrastructure can move those requests, the more of them -- and in this case the more virtual machines -- it can support.
While 8 Gbps Fibre Channel will eventually be the same price as today's 4 Gbps Fibre Channel, SSD and server virtualization mean that more customers will need to move to a faster protocol before price erosion occurs. This is an ideal area for a storage solution provider to add value. You can provide guidance in determining if your customer has a need for that faster performance, and you can provide advice on whether they move to 8 G bps Fibre Channel or another, faster protocol.
In today's infrastructure, "faster" essentially means 8 Gbps Fibre Channel, 10 Gigabit Ethernet or Fibre Channel over Ethernet (FCoE), which operates over 10 Gigabit Ethernet. For many environments, 8 Gbps FC will be the path of least resistance since it doesn't involve a new protocol to learn and integrate. IT organizations can upgrade the parts of their infrastructure that can take advantage of the higher-performance 8 Gbps FC and leave the rest untouched. You'd simply install 8 Gbps cards into the servers and integrate 8 Gbps switches or, if your customer has a backbone/director architecture, sell the 8 Gbps blades. That part of the environment simply becomes faster.
VARs always have to balance all the new stuff they are learning about against what their customers are ready to and can afford to invest in. The advantage of being able to upgrade specific hot spots in the environment without having to learn something new has particular value in a tough economy. Budgets are stretched thin, and IT staffs are stretched even thinner. Incremental upgrades that simply plug in and act like the old components they replaced (except that they're faster) will be very popular.
Beyond that advantage, a case can be made that 8 Gbps Fibre Channel is actually faster than its 10 Gigabit counterparts. 10 Gigabit IP and iSCSI will both have IP overhead that they need to work through. As a result, the real-world throughput of 10 Gigabit IP-based protocols is rarely full line speed. While there are iSCSI host bus adapters (HBAs) that can handle offloading IP, this will significantly add to the cost of the iSCSI deployment. Even FCoE may have to carry IP traffic along the same segment as the Fibre Channel traffic, resulting in less bandwidth for the Fibre Channel part of the environment.
Finally, one other consideration that may come up as an objection to 8 Gbps Fibre Channel: an uncertain upgrade path beyond 8 Gbps. But if you move a customer to 8 Gbps FC now, there is nothing that would preclude them from moving to FCoE in the future. Nor is there reason to believe that 8 Gbps is Fibre Channel's last rodeo. There has already been commitment from several vendors for support of 16 Gbps Fibre Channel as soon as next year. Moving a customer to 8 Gbps Fibre Channel now will most likely provide them with the same upgrade or cross-grade options as if they wait. But waiting could be problematic. They could very easily sacrifice performance, which typically costs an organization time, money or productivity -- and, in some cases, all three.
About the author
George Crump is president and founder of Storage Switzerland, an IT analyst firm focused on the storage and virtualization segments. With 25 years of experience designing storage solutions for data centers across the United States, he has seen the birth of such technologies as RAID, NAS and SAN. Prior to founding Storage Switzerland, George was chief technology officer at one of the nation's largest storage integrators, where he was in charge of technology testing, integration and product selection. Find Storage Switzerland's disclosure statement here.
This was first published in December 2009