Opinion: The Real SMI-S Agenda

Someone just sent me a copy of a new "white paper" entitled "Customer Reaction to SMI-S." With due respect to the paper's author, I have to say that it left me a little perturbed. I worry that it might strike storage technology consumers the same way--a concern only because I rather like the author's storage company, which not too long ago was just a little start-up with big ideas.

The first reason I took exception to the piece is that it gave the impression that the Storage Networking Industry Association's Storage Management Initiative-Specification (SMI-S) was a sort of inevitable thing: a storage management panacea that every vendor and consumer would embrace going forward. I found myself wondering why a writer from a newbie storage vendor would suddenly be carrying the mantle for SNIA, an organization hardly known for its warm embrace of start-ups according to many of the innovators I've had the pleasure to interview over the years.

A quick visit to the author's web site gave me a clue. Last November, at the Fall Storage Networking World show, the author's company had announced its products' SMI-S interoperability. That seemed to be at least a partial explanation. You don't need to be Sun Tzu or Machiavelli to realize that, in order to get in good with the SNIA ruling clique, you at least need to talk the talk of SMI-S--and keep your membership dues up to date, of course.

What also struck me about the paper was the bell curve used to describe SMI-S adoption by the user community. It seemed that, in addition to adopting the rhetoric of SNIA about SMI-S, the author was also borrowing a page from Gartner's book.

Industry research and analysis firms are fond of presenting their paying clientele with evolutionary curves that show how the client's technology is moving from early adopter to midstream to late adopter phases. It's comfort food, but without any substantive nutritional value, and it keeps the client happy. In essence, the evolutionary curve says that the client's technology is going places, that its marketing efforts are succeeding in overcoming the fear, uncertainty and doubt among conservative consumers, and, best of all, that the technology will soon go mainstream, yielding big profits for its creator.I couldn't help thinking about other technologies that got the "evolutionary inevitability" treatment from Gartner, IDC, and the other research houses over the past few years. Remember how network computers were going to replace PCs? Remember how application service providers (ASPs) were going to convert software into services? The same industry analysts showed similar evolutionary curves to vendors of those products (and to their investors and stockholders), projecting big wins that somehow never materialized.

Notable in its absence from this white paper was any formal demand-side data to support the trend chart. A sprinkling of unattributed quotes, presumably from customers, does not substitute for hard statistical data. In the author's defense, he says that he was communicating sentiments expressed to him by a few of his customers. But how one could extrapolate a wave of support for SMI-S from a few casual statements eludes me.

The problem is that hard data don't exist. That's because no scientific surveys have been made to gauge user views or timetables for SMI-S adoption. This, in turn, is because the bulk of storage consumers have never heard of SMI-S. And that's because most vendors have not implemented the quasi-standard (SNIA is not a standards group) on their gear and because SNIA has been very ineffective at customer outreach.

SMI-S has been adopted by most of the three-letter, big iron, storage array vendors in large part to offset customer perception of their proprietary ways, but also to differentiate their wares from those of smaller innovators. Some would argue that an unstated goal of SMI-S is to keep the small fry out of their market by elevating SMI-S support to the status of a checklist item in users' requests for information (RFIs). This effectively freezes out the products of the little vendors, who arguably make better gear than their larger cousins. Often, however, they lack the funds to implement SMI-S providers and note that customers aren't willing to pay extra for SMI-S support in any case.

I can't help but notice that SMI-S is being pushed by the very same vendors who just a couple of months ago were arguing that their logical unit numbers (LUNs) were different from their competitors' LUNs, effectively nixing any chance of a workable virtualization scheme. They are also the same group who engineered Fibre Channel standards to create products that would comply with the letter of the standard and absolutely would not work with their competitors' standards-compliant products.Would the white paper author, or the SNIA, expect us to believe that there will be a different outcome from the implementation choices made by these same vendors?

I have also noticed that most of the current crowd of SMI-S "enabled" hardware vendors are also UNIX/RISC/LINUX bigots, which might help to explain why Microsoft bailed out of SMI-S development early on. It remains to be seen how much sway SMI-S compliance will have with customers operating in a predominantly Windows-centric world.

Another reason for the lack of awareness of the SMI-S spec among users is SNIA's generally poor job of ombudsmanship. This is a matter of record and is reflected in the many "educational sessions" conducted by the organization that quickly descended into lackluster discussions of meta-modeling theory and object oriented programming. Does meta-modeling theory resonate with storage customers? Ask those who have attended the sessions, once they rub the sleep from their eyes and down a couple of cups of coffee.

Don't get me wrong. I believe we need some sort of overarching storage management scheme to drive cost and complexity out of storage. What I find really difficult to believe is that the very same vendors who, through their disdain for their customers' needs and myopic focus on their own profit margins, first introduced proprietary barriers to interoperability, cross-platform virtualization and storage management, are now suddenly going to work together to subjugate their wares to a common management scheme.

Are we really expected to buy into the proposition that big iron vendors want their wares to be treated as "peer objects" in an all-inclusive management meta model? Cross-platform management, like standards generally, exerts a commoditizing influence on technology, if not in reality, certainly in terms of customer perception. One would have to be naive, to use a kinder, gentler tone favored by three-letter storage vendors these days, to find the thesis of this paper credible.I, for one, am not convinced about SMI-S. And I doubt that many of you readers of this column are either.

In deference to the white paper author, I contacted him to discuss my concerns. From this brief interview, two additional points seemed worth reporting.

First, the author went to some pain to inform me that his company's products were designed from the outset with SMI-S support inside. Thus, it would be incorrect to characterize him or his company, he argued, as turncoats who suddenly became shills for the SNIA's marketecture.

Secondly, he noted for the record that his SMI-S implementation was intended to provide simplified management across his own extensible equipment, which, he said, is hailed by those who use it as the most user-friendly and hassle-free storage around.

Thus, in this context, SMI-S has been implemented, not to promote smooth control of heterogeneous storage environments, but to serve a particular vendor's quest to replace heterogeneous infrastructure with its own homogeneous gear. This sounds vaguely familiar to pitches I have heard from other vendors of monolithic storage solutions.To his company's credit, said the author, its SMI-S implementation comes free with the gear, while a lot of three-letter array vendors charge extra for their implementation.

All of which goes back to the point: Is SMI-S the promised management panacea for the heterogeneous storage infrastructure, or just another point product designed to join us all at the hip to one vendor's wares? Drop me a line