Almost four month ago Brocade announced a plan to begin selling Fibre Channel adaptor cards or HBAs in June this year, using technology supplied by LSI.

But the very mature HBA market has long been heavily dominated by QLogic and Emulex, and HBAs like other Fibre Channel products require an expensive amount of third-party certification testing.

For those reasons analysts from both the IT industry and Wall Street were very doubtful of the wisdom of the move.

It comes up in pretty much every call: What were you guys thinking?, Brocade CEO Mike Klayko said candidly at its investors’ conference in Boston, Massachusetts, last week.

Brocade’s answer is principally that it will meet future needs for end-to-end storage networking services far better than Emulex or QLogic, because unlike them it owns the rest of the storage network. Emulex and QLogic will have to find partners. From a fabric model that will be much harder for them. It’s much easier if you own the pieces from a performance and testing point-of-view, Klayko said.

But Brocade did not detail exactly what features will be unique to its HBAs, or say when they will ship.

Brocade’s product development vice president Don Jaworski said that customers moving applications and virtual servers between physical machines need a SAN that can keep up with those moves. But he also acknowledged that QLogic and Emulex will embrace the virtualized world.

The legacy HBA makers [QLogic and Emulex] are participating in the virtualized architecture. We’ll do that too. That’s table stakes, he said.

But server virtualization will create a need for features that are not currently supported by HBAs, he said. Take something like QoS. All of a sudden you’re going to need QoS from end-to-end. If I’m an HBA vendor, I can’t really figure out how to do that. We’ll drive that, because we own the pieces.

Enterprise Strategy Group analyst Tony Asora said that Jaworski’s references to data movements and policy-based management at the server level were interesting, and right-thinking, but noted that Brocade made no commitments about what it will develop.

The HBA market is tough. Not so much the technology but the testing and interoperability. It is a huge and expensive undertaking. Do I think it’s a wise move for Brocade? No. But if they do actually elevate the role of the HBA it could be brilliant, Asora said.

Ovum analyst Carl Greiner was much more skeptical. The future sounds nice, but they’ve got to get the server makers of the world buy into it, and into a huge amount of testing and integration. And I don’t believe that Emulex and QLogic are going to stand still, Greiner said. His final shot was: Does Cisco make Ethernet cards? No, it doesn’t.

Illuminata analyst John Webster said that in June he had trouble understanding Brocade’s HBA plan. Since then he has changed his view.

When I saw Cisco’s VFrame. It became clear why Brocade wants to do this, Webster said. Launched this year, the VFrame was launched is an appliance that runs virtualizes Ethernet and Fibre Channel networks, by automating their configuration and re-configuration.

Think about the end-points. It’s not clear to me what Cisco’s end-points are, or that it really has a view of what’s going on in the server. You need that view if you’re going to go end-to-end, he said.

Webster also backed Brocade’s claim that because HBAs are such a frequent cause of SAN problems, end users will want to buy Brocade-branded HBAs. I can see at least a percentage of them saying give me a Brocade HBA, and take that headache away, he said.

On the overall topic of SAN developments, Webster said: There’s an awful lot of white spaces in what Cisco and Brocade say about this topic. They’re in a race, but nobody knows what the finish line looks like.

Sam Wilson, managing director at JMP Securities said that Brocade has a reason to move into HBAs quite separate and over and above network virtualization.

It’s about bundling, and the fact that Cisco doesn’t sell HBAs, he said. Low-end and mid-range customers will like being able to buy HBA and switch bundles. This isn’t a profit or revenue move, but a way for Brocade to sell more switches, he said.

Our View

The debate about Brocade’s HBA move illustrates the need to virtualize networks in order to match server virtualization. And if Brocade is telling the truth when it says that it will be much easier to do that if it owns all the components, then it also highlights the relationship between technology convergence and consolidation among IT suppliers.

One side of the deal that has not been discussed is LSI’s role. LSI has invested plenty of money developing HBAs, but has never been able to claw anything more than a tiny slice of the market away from Emulex or QLogic .

That gives LSI a motive to offer Brocade a very good deal on HBAs, which in turn could allow Brocade to offer server makers and other resellers very good terms perhaps even upfront payments to help cover their annual certification and testing costs.