IBM and FCoE: The Unified Networking Superstore

I expected this series about vendor perspectives on FCoE to wrap-up last week, but a few very prominent industry leaders have stepped up and asked to be heard. One of those companies qualifies as prominent because of its extensive experience (founded in 1896 as the Tabulating Machine Company), track record of innovation (more than 40,000 active patents), and proven results in bringing information technology to market (over $100 billion in revenue). The company is also known affectionately as Big Blue.

Responding to my questions for IBM was an individual equally impressive in the field of storage networking. Clod Barrera is a Distinguished Engineer and the Chief Technical Strategist for IBM System Storage, a founding member of the Storage Networking Industry Association and has served on its Board of Directors.

We started by asking if Fibre Channel over Ethernet will cause the storage industry to finally converge on Ethernet. Barrera acknowledged that FCoE is the most promising unified networking technology IBM has seen in years. But he stopped short of endorsing FCoE as the certain platform for convergence. A veteran of multiple technology transitions, Barrera provided specific examples of continued momentum for Fibre Channel and iSCSI that will prolong the transition and perhaps give iSCSI the head start it needs to be "the" platform for the convergence of storage on Ethernet.

Example #1: A year and a half after 8-Gbps Fibre Channel networks hit the market, 8-Gbps FC arrays are just now being offered. It will take another two years plus for 8-Gbps FC storage to displace 4-Gbps FC storage. That's a four year adoption cycle and puts us into 2011. IBM expects meaningful FCoE adoption to take even longer, allowing enough time for yet another generation of 16-Gbps Fibre Channel networks and storage. If 16-Gbps networking hits the market as early as 2010 and has the same four year adoption cycle, 16-Gbps Fibre Channel will offer a 60 pecent performance advantage over 10-Gbps Ethernet through 2014. I would expect this to delay the emergence of native 10Gb FCoE storage.

Example #2: Let's assume that production class 10-Gbps FCoE networking is available at the end of 2009. Then let's apply the year and a half that 8-Gbps storage lagged 8-Gbps networks to native 10-Gbps FCoE storage. That puts us into 2011 for a complete FCoE SAN. Given the momentum that iSCSI has, I believe that by 2011 it's entirely possible iSCSI could achieve a commanding share of storage, switch and host ports vs. Fibre Channel and FCoE. And in my opinion, if iSCSI is able to grab over 50 percent market share, there may be no turning back. Customers will have the Ethernet-based unified networking they want and FCoE from then on serves as a bridge to what becomes legacy Fibre Channel and FCoE storage.

As for IBM's position in the transition, Barrera explained that IBM is in a great position because the company is the ultimate solution supplier. I agree. IBM is a unified networking superstore that carries a highly diversified range of very complex technology. And regardless if the company invents the technology or sources the technology from another vendor, IBM is the absolute best at integrating best-of-breed technology with world-class software and services to provide infrastructure tailored to support a business.
Barrera says IBM believes that FCoE networking will become mainstream in 2010 as end-users take advantage of the fact they can keep deploying familiar iSCSI or Fibre Channel storage, but reduce cabling and the number of host adapters they need by using Converged Network Adapters. Storage is another matter. IBM expects multiple technologies to share the market through 2013.

I have to agree -- Storage trends always take longer than you think...

We're still looking for more tech vendors to provide their ideas and insights on the forthcoming technology transition, so get in touch.