After the close of trading yesterday, Brocade announced that it was going to buy Foundry Networks for $19.75 in cash and stock or a total of $3 billion dollars. This acquisition puts Brocade in a much better position in the coming data center network wars, as just being the dominant Fibre Channel switch vendor isn't worth much as large enterprise data centers move from separate storage and communications networks to a converged Ethernet.As the FCOE advocates see it, each server needs just 2 10-Gbps Ethernet cables to a pair of "top of rack" switches, and those two connections will carry storage and other network traffic like peaches and cream across a converged network. Since Cisco is the only vendor to make both Fibre Channel and Ethernet switches, it has been a big proponent of FCOE (Fibre Channel Over Ethernet) and has the first, pre-standard, top of stack switch.

For Brocade to remain a player in the data center, it needs to have more than 10-Gbps Ethernet blades for its DMX switch or Cisco will just keep nibbling up the edges. The network guys will use Cisco switches for the servers and just uplink them to the DMX for storage services. Once that happens, Cisco looks like a natural when the time for a new core switch comes around. With the Foundry buy, Brocade picks up a respectable line of high-end Ethernet switching gear it can use to maintain its position as FCOE comes to market

I've written about how I dislike FCOE in the past, even called it a protocol nobody needs, but in my old age I'm starting to mellow. Running Fibre Channel Protocol over 10-Gbps Ethernet lets those large organizations with well-managed Fiber Channel infrastructures use the hard-earned knowledge and tools they use to manage the FC network to run the FCOE net.

One of the reasons I've been doubtful about FCOE is the question of ownership of the converged switches FCOE travels over, especially the combination switches that have both FC and 10-Gig ports. The network group in most organizations is going to be yelling "They're Cisco Ethernet switches so we should run them," while the storage admins scream "You guys don't understand storage and will bankrupt the company if we let you run the SAN switches."

Brocade stands to do well in the organizations where the storage guys lead the way. Due to the sales process, finger-pointing, and a history, now mostly resolved, of Fibre Channel interoperability issues, the vast majority of Fibre Channel SANs are more homogeneous than a typical LAN with one vendor's switches and HBAs predominant. SAN guys buy their switches from their storage vendors (more than 90% of Brocades sales are OEM through EMC, HDS, and the other array vendors), so Brocade's relationships with the OEMs should help it compete against the ubiquitous Cisco rep.

Clearly, Brocade had to sell Ethernet, or more accurately Converged Enhanced Ethernet (CEE) that Cisco calls Data Center Ethernet, the lossless, prioritized super-duper Ethernet variant that FCOE really runs on. Foundry was the best-positioned of the remaining players to remain a going business while also supporting Brocade's needs in the data center. It could have bought Force10 Networks, Extreme Networks, or Woven Systems for less, but it would have gotten less.

As a Foundry customer at the college, I'm going to have to see how things shake out. Brocade may decide to end some of Foundry's peripheral product lines like server load balancers and wireless. For now, I'm just glad my sales rep, who was our SE for years, can cash in his options.