Network Computing is part of the Informa Tech Division of Informa PLC

This site is operated by a business or businesses owned by Informa PLC and all copyright resides with them. Informa PLC's registered office is 5 Howick Place, London SW1P 1WG. Registered in England and Wales. Number 8860726.

The FCoE Effect

In a recent cover story for Information Week on the Unification of the Data Center Infrastructure, I discussed Fibre Channel over Ethernet (FCoE), Converged Ethernet (CE), and related new technologies that will have an impact on data centers. An area that I had not delved into until posed the question by a Byte and Switch reader was, "How will this impact the Host Bus Adapter and Network Interface Card vendors? Will they go away?"

The concern is that if the value proposition of FCoE -- reducing the number of network and storage I/O cards in a server -- is realized, then would this also reduce the amount of these types of vendors? The big assumption here is an FCoE card will cost the same as a HBA card does today. I don't think that is likely to happen very soon. Thus far, FCoE cards have been significantly more expensive than their conventional equivalents because you are getting essentially two cards for that price (IP and FC). So while there may be possibly fewer cards sold, they will sell at a higher price point -- offsetting the loss in units with more revenue per unit.

If FCoE is successful -- and that remains a big "if" for now -- my expectation is that the big winners will be the storage HBA vendors like Brocade, Emulex, LSI, and QLogic, and the traditional IP card manufacturers could be at risk. A big component of the unified infrastructure is the transport of the storage protocol, and it seems likely that you will want to buy those cards from a storage vendor. As a result, IP comes along for the ride.

While there are a few traditional IP NIC card manufacturers investing in FCoE, most of them, like Neterion and SolarFlare, are focused on making better IP cards by adding capabilities like Quality of Service and channeling to be able to divide up the bandwidth. These companies are pushing harder into faster IP speeds like 40 Gbit/s and 100 Gbit/s.

Even if FCoE is successful, it seems that it will be so more on the storage side than the IP side. I think there will always be certain servers in the environment that will have a specialized and high-speed NIC, which of course defeats the whole purpose of a unified infrastructure.

  • 1