It seems I made an error in a blog a few weeks ago about Fibre Channel, although it doesn't change my view about the long-term viability of standalone FC storage networking.
In this post on March 13, I said that the Fibre Channel industry was only just beginning the conversation on 16 Gbps Fibre Channel, and that it would be quite a while before we see any actual technology available. Well, this week at the Optical Fiber Conference and National Fibre Optics Engineers Conference (OFC/NFOEC), a number of companies unveiled production models of 16 G Fibre Channel components
Gennum Corp., for example, is offering immediate availability of its new IC that puts 16 Gbps Fibre Channel, along with clock and data recovery (CDR), integrated limiting and transimpedance amplifiers and an EQ/laser driver, on an SFP+ form factor. The device also uses the same reference design as current 8 Gbps solutions, allowing for an easy migration to the higher data rate, currently expected to be 14.025 Gbps when the International Committee for Information Technology Standards (INCITS) finishes the standard later this year.
At the same show (which is probably the most exciting time of the year for network engineers) Finisar demonstrated a range of new optical systems, including a 16 Gbps FC transceiver that can operate across 150-meter lengths of OM3 fiber. Also built on the SFP+ form factor, the device uses shortwave signals designed for rock-solid reliability and line integrity for enterprise storage applications.
Also on tap was Opnext, which has developed a long-wave transceiver for long-haul operations. The unit uses 1310 nm DFB lasers capable of maintaining a 16 G link across 10 km or more. The device opens up the possibility of not only centralizing storage on the home office LAN, but across the wide area as well.
Clearly, 16 Gbps Fibre Channel technology is much farther along than I thought, but I still stand by the point of my original post: The future of independent Fibre Channel networking hardware is still very much in doubt. Fibre Channel over Ethernet makes it possible to run the protocol as an Ethernet layer, which likely will be cheaper and more flexible than a discrete FC infrastructure.