The powers that be in the Fibre Channel world are drinking champagne to celebrate that they've dotted all the Is and crossed all the Ts on the ANSI INCITS T11 standard for 16Gbps Fibre Channel. The usual suspects (Brocade, Emulex and Qlogic) have all announced enthusiastic support for the new faster standard, but given the pattern of the past few upgrades, I'm not expecting customers to beat a path to their doors.
In the past, customers were chomping at the bit to get faster Fibre Channel products as soon as the vendors released them. The 1-2Gbps and 2-4Gbps transitions went pretty quickly as per-port costs for electronics and optics were pretty much the same and enterprises were still expanding their SANs to include more servers. After all, if a 2Gbps HBA (Host Bus Adapter) or switch and the 4Gbps equivalent cost the same and are interoperable, why buy the slower one?
The 4Gbps to 8Gbps transition has been a bit rockier. 8Gbps optics have been more expensive than their 4Gbps equivalents, and customers haven't been buying 8Gbps HBAs in droves. Today, two years after 8Gbps HBAs hit the streets, most vendors are still selling at least as may 4Gbps HBAs as 8Gbps ones.
We can only expect that the cost difference between 8Gbps and 16Gbps Fibre Channel devices and their 8Gbps equivalents will be even bigger. The 16Gbps uses a more efficient coding scheme than slower FC implementations, so 4/8/16Gbps ports will need to implement the new 64/66 bit encoding as well as 8/10 bit. Add in that 16Gbps SFP+ optics need to implement a re-clocking circuit and we're talking about significant additional complexity that will probably equate to greater cost.
Then there's the pressure from the various 10Gbps Ethernet storage technologies. Smaller organizations that may have implemented Fibre Channel in the past can use NFS or iSCSI over 10Gbps Ethernet while large enterprises can use FCoE and retain the Fibre Channel management tools they've been using. These factors, along with server virtualization, will reduce Fibre Channel port sales, leaving a smaller number of ports across which FC vendors must amortize their development costs.Howard Marks is founder and chief scientist at Deepstorage LLC, a storage consultancy and independent test lab based in Santa Fe, N.M. and concentrating on storage and data center networking. In more than 25 years of consulting, Marks has designed and implemented storage ... View Full Bio