Time for a management standards 'reality check'

By Heidi Biggar

At the Storage Networking World show in October, 19 vendors banded together to show end users that it is possible to manage a multi-vendor storage area network (SAN) using a common interoperable interface, or backbone, based on the Storage Management Initiative Specification (SMIS), formerly known as Bluefin.

In the interoperability demonstration, 32 products were tested—including disk arrays, tape libraries, switches, host bus adapters, network-attached storage (NAS) appliances, virtualization products, and storage management software—and 93 points of interoperability were established using the SMIS backbone.

Participating vendors included AppIQ, BMC, Brocade, Computer Associates, Crossroads, EMC, Hitachi Data Systems, Hewlett-Packard, IBM, Inrange, InterSAN, LSI Logic, McData, Network Appliance, Prisa Networks, QLogic, Quantum, Sun, and Veritas.

Click here to enlarge image

But while the interoperability demonstration served as a proving ground for basic standards-based management (it demonstrated the use of Bluefin technology as a standards-based API and CIM as the underlying object model for building multi-vendor management interoperability), some say it is still far too early for end users to begin factoring SMIS support into purchasing decisions.

According to Phil Goodwin, program director for server infrastructure strategies at the META Group, SMIS support shouldn't be factored into the purchasing equation for at least 18 months, at which time SMIS-compliant products will be more widely available and more mature.

"We'll reach a point when the technology has matured and there is minimal difference among vendor products," explains Goodwin. At that time, it may make sense to include standards support in the vendor-selection process, he says.

In the meantime, Goodwin cautions end users against putting off storage purchases because of the specification. "IT organizations that wait for standards-compliant software will delay improvements to their own storage management systems, to their own detriment," he says.

So, when will end users see widespread SMIS support among storage vendors? The Storage Networking Industry Association (SNIA) says it expects 60% compliancy among participating SMI Forum members by 2004, and 100% compliancy by 2005.

Goodwin expects to see a broad range of CIM- and SMIS-compliant products in end-user production in the 2004-5 timeframe, but says end users shouldn't expect any real value-add from the specification for at least three years.

"Early implementations will be limited to simple device discovery, configuration [e.g., LUNs and data zones], and error management," explains Goodwin in a recent META report. "More advanced functionality [e.g., local snapshot replication] will not be offered using CIM specifications before 2006-7."

Goodwin says he expects about 80% to 90% of vendor device functionality to be included in the SMI specification by 2007.

Formed by SNIA last month, the SMI Forum is charged with the responsibility of driving adoption of standards for multi-vendor interoperability. The forum will initially focus on promoting the SMI specification. More than 25 SNIA member companies have indicated their intent to participate in the SMI Forum.

The SMI Forum is not to be confused with a separate effort by SNIA members Hitachi Data Systems, IBM, Sun, and Veritas.

In mid-October, these four companies announced a non-exclusive partnership to promote the SMI specification, which hinges on a mutual commitment to deliver standards-compliant products in 2003, to support the emerging SMI specification and CIM/WBEM interfaces, and to jointly test and qualify products by making SMI agents available to participating companies for testing.

The reality is that for the next several years the SMI specification will have little, if any, consequence for end users. Why? Because vendors—despite what they may say—simply have too much to lose at this time to share intellectual property with competitors for the purpose of standards development.

"As with device discovery, critical functionality will not be delivered using standards until the commercial value has already been extracted from the product and vendors have moved on to new categories [e.g., storage virtualization, real-time backup and recovery, and chargeback]," Goodwin says.

The point here isn't that the SMI specification is without purpose, but that it is not as meaningful as vendors say it is—at least not yet.

"The longer-term IT benefit will be the ability for a single management product to perform advanced management functions for heterogeneous storage hardware," says Goodwin. "However, this functionality will be available from vendor-specific initiatives and 'consortium/partnership' API exchanges."

Consider the recent cross-licensing/API agreements between Hitachi and IBM, Hitachi and Hewlett-Packard, HP and IBM, and HP and EMC—all of which were done against a backdrop of active standards development (see "HP swaps APIs with IBM, Hitachi," InfoStor, October 2002, p.8).

Such agreements are designed to provide end users with an interim way of managing multi-vendor SANs while standards are being developed, integrated into products, and made available to the end-user community.

The problem with these types of agreements is that they are limited in terms of breadth and scope.

The agreements extend to only certain vendors and the technology swap is typically such that they allow competitive vendors to "see" inside the box, but that's about it.

These agreements do nothing to promote unified management of multi-vendor SANs or to lower the cost of SANs—two major impediments to widespread SAN adoption. And it is for this reason that hardware, software, and component vendors—in the context of SNIA—are pushing standards development with an all-out media blitz.

"The [unspoken] core mission of standards bodies is to promote the adoption of member products," says Goodwin. "This leaves IT organizations with the difficult challenge of differentiating between what is useful technology and what is hype designed to soothe a hesitant market. Right now, we believe SMIS currently falls into the latter category."

This article was originally published on December 01, 2002