The Common Code

How important are standards? This seems like a simple question (or perhaps even a simpleton's question). Of course, standards are very important, you might say. It's like asking whether education is important. Or water. Or beer.

On an abstract level, increasing the interoperability among multiple vendors' storage systems and devices is unquestionably – A Good Thing. Customers don't have to be fearful of being locked into proprietary technology, thus benefiting all vendors and (theoretically) expanding the total market.

The Storage Networking Industry Association (SNIA) has been driving these points since its inception, and the group has set as its goal having all storage managed by the Storage Management Interface Specification (SMIS) by 2005. SMIS is based on years of standards work in the network management area, primarily the Distributed Management Task Force's Common Information Model (CIM) (see SNIA Puts the Pieces Together, SNIA Forms Management Forum, and Bluefin Swims to the Surface).

This is both an ambitious and laudable vision. But there's built-in resistance that every standards movement faces. It boils down to two things:

Listen, for example, to John Cloyd, who heads up the software business of Overland Storage Inc. (Nasdaq: OVRL), which sells storage resource management (SRM) software. I'm not trying to single John out; he just happens to be somebody I was recently talking to about this.

From Overland's perspective, are storage management standards important? "When there's a standard, we'll support it," says Cloyd. But why not support SNIA's current SMIS? "We're not losing any deals right now because we don't support standards," he says. "What's winning us deals is the feature-set differentiation."

Refreshingly honest. But this is precisely the kind of inertia that keeps standards from being widely used and accepted.

A dirty little secret of the storage management standards effort is that it's forcing some vendors to overhaul their applications. This is an expensive, time-consuming ordeal that won't add to the bottom line anytime soon. We all want a lingua franca. But no one has an incentive to fund its development in the near term.

Yes, four vendors – Hitachi Data Systems (HDS), IBM Corp. (NYSE: IBM), Sun Microsystems Inc. (Nasdaq: SUNW), and Veritas Software Corp. (Nasdaq: VRTS) – have publicly committed to delivering CIM-based products this year. You have to start somewhere. But will these "standards-based" products do anything useful? (See Four Vendors Back CIM.)The issue here is the complex process of deciding what should be part of the baseline standard. Gary Gysin, VP and general manager of McData Corp.'s (Nasdaq: MCDTA) SANavigator software division, says the functionality in the SNIA spec just isn't there yet. "If CIM was fully baked with 100 percent of the functionality we need, we'd be supporting it today," he says. "The question is: Does CIM expose enough? Is it full-featured enough yet?"

Well... no, apparently, not just yet. Until it is, there will continue to be the need for various API swaps and vendor-specific development (see HP Makes API Triple Play and EMC Swaps APIs With Brocade).

The point is that the SNIA-led standardization efforts won't bear fruit until there's a critical mass of real products that do things that are actually useful. This will happen, surely. But this year? I doubt it. Maybe next year? Or 2005? We'll see.

And there's that other sticking point – the concern among vendors that CIM could eliminate certain software-licensing revenue streams. But this is unfounded, says A. "Ash" Ashutosh, CTO and co-founder of AppIQ Corp., a developer of CIM-based SAN management software. He believes device-specific software will never go away.

"EMC Corp. [NYSE: EMC] can never manage a Hewlett-Packard Co. [NYSE: HPQ] box better than HP can," he says.The idea is for SMIS to handle, say, 90 percent of the tasks a user needs to perform. What will happen, Ash predicts, is that storage hardware vendors will extend CIM to handle device-specific functions in their own software. "That will let them keep that value-add for themselves and not lose on the software revenues." He adds, however, "I think the value of those licenses will start to come down, because now you're not using that software 90 percent of the time."

Let's take a look at what really got the ball rolling on the storage management standardization front. The approximate date it kicked into high gear was Oct. 29, 2001, when EMC introduced WideSky – a middleware layer that supposedly would allow EMC to manage any other vendor's storage (see EMC Automates).

WideSky made every other storage vendor freak out. They envisioned (perhaps rightly so) a pincer move by EMC to control the storage software market.

"EMC, to its credit, listened to the market very closely and saw the need for standardizing all storage management," says Ashutosh. "But frankly, none of the other guys wanted to make EMC succeed again."

The repellent aftertaste of WideSky prompted a coalition of vendors – that same gang of four again: HDS, IBM, Sun, and Veritas – to use the CIM standards effort as a competitive marketing cudgel against EMC. Perhaps we should give them the benefit of the doubt and assume their pressure forced EMC, at least nominally, to pledge full support of the SNIA standards. By contrast, less than a year ago former EMC CTO Jim Rothnie declared: "If we wait for standards, they'll be ready... in 20 years' time." Not exactly the words of someone interested in joining a standards-setting effort (see EMC Outlines CIM Support Plans, Standards Clique Freezes Out EMC, and Standards Battle Still a Hot Potato).Again, enterprise users of storage systems understand the value of standards. They're certainly not interested in getting locked in to a single vendor. But the evidence suggests that the market, to some extent, must be dragged kicking and screaming toward standards. SNIA is in a position to do this, but it needs the commitment of its members, too.

James Staten, director of software marketing in Sun's network storage division, has pointed out that TCP/IP – as a vendor-neutral technology – needed to be given a push before the Internet could blossom. Staten, who was speaking at the RBC Capital Markets