SMI-S Leaves the Starting Line

At the Storage Networking World Conference in Phoenix last month, the Storage Networking Industry Association (SNIA) announced that more than 100 products from 14 vendors had passed its testing program and conformed to the Storage Management Initiative Specification (SMI-S). Industry leaders such as Brocade Communications Systems, EMC, Hitachi Data Systems, HP, and IBM all have products certified as conforming to SMI-S. Moreover, SMI-S 1.0.2 has entered the InterNational Committee for Information Technology Standards (INCITS) Fast Track process and could become an ANSI standard as early as this summer. Although both of these announcements represent a major milestone, SMI-S's promise of simplifying management of heterogeneous storage networks, potentially from a single point of control, is still relatively far off--if indeed it will ever be realized.

SNIA has been working on the SMI-S since at least 2002. In broad strokes, the standard's objective is to develop and standardize interoperable storage management technologies, thus letting developers more easily support products from different vendors. Moreover, by providing a common management interface for network components, SMI-S could dramatically reduce the complexity of managing heterogeneous storage networks. If the vision is realized, introducing a product that conforms to SMI-S into a Storage Area Network (SAN) will signal the product's presence and functions to the other network components, and the product will be able to share its resources appropriately.

SMI-S rests on several factors that bode well for its eventual adoption. First, it's based on Web-Based Enterprise Management (WBEM), a broad set of management and Internet standard technologies developed by the well respected Distributed Management Task Force (DMTF) to unify management of enterprise computing environments. WBEM in turn embraces the Common Information Model (CIM), a language and methodology for describing management data. The bottom line is that SMI-S neatly integrates into related initiatives.

Second, the standard will offer benefits to both vendors and end users. Vendors will no longer have to invest time and money to develop and integrate interfaces from multiple suppliers into their products. By using a common, standard interface, vendors can concentrate instead on adding features and functionality that will deliver real value to their end users. In addition to being able to bring better, more innovative products to market faster, vendors will have a broader market to address because all products that conform to SMI-S will be able to work with all other conforming products.

The advantages to end users are clear as well. SMI-S should dramatically reduce the complexity of simultaneously managing a lot of different SANs. Currently, because enterprises typically run multiple OSs and different applications on each OS, most enterprises have no choice but to integrate storage networks from multiple vendors. Claims of interoperability are based on specific vendors cutting deals to swap specifications for their APIs. But as Jamie Gruener, a senior analyst at the Yankee Group, pointed out in a recent Webcast, managing heterogeneous SANs frequently just doesn't work well. End users are confused and have discounted many vendors' interoperability claims. SMI-S could not only reduce the complexity of managing different SANs, but could offer end users a certain degree of confidence that products will, in fact, interoperate.

A standard interface also offers end users other benefits. Companies don't have to lock into a single vendor and can instead work with best-of-breed point solutions, and deploying new technology should be easier. Finally, more efficient network management could reduce downtime.

Consequently, it isn't surprising that despite the difficulties to date, enterprises are still very interested in consolidating storage-network management. According to a survey of storage decision makers by the Yankee Group, last year 37 percent either implemented or planned centralized storage management solutions. Another 35 percent indicated that a single management approach for storage was a high priority.

But despite all this momentum behind SMI-S, doubt remains about whether the standard will work. Some observers believe that vendors aren't entirely committed to SMI-S. Moreover, the steps that an enterprise will have to take to move from legacy systems to a system that conforms to SMI-S aren't entirely clear. Also unclear is how API-swapping programs, which are the basis for interoperability now, will integrate with SMI-S. In short, although SMI-S has left the starting line, the standard has a long way to go to reach the finish line.