In early 2002, a consortium of storage vendors created a new SAN management initiative called Bluefin. Compared to previous management schemes, this collective effort was unique in its focus on creating common APIs that could span vendor lines.
In addition, it would establish a new interface based on CIM/WBEM, using the common information model object-oriented standard with Web-based management platforms.
By the end of 2002, Bluefin was brought into the Storage Networking Industry Association (SNIA) for further work and standardization in the form of the storage management initiative (SMI).
By 2004, SMI 1.0 was formulated by the SNIA and received official standardization via ANSI/INCITS 388-2004. Promoted as a final answer to the pent-up frustration of storage administrators, SMI emerged more than a decade after the original creation of Fibre Channel SAN standards and products, and well after the proliferation of SAN technology throughout all major institutions and enterprises worldwide.
Hardware vs. SMI
Obviously, the development and deployment of storage networking technology has far outpaced our ability to readily manage storage infrastructures. Why? In large part, the lack of comprehensive management tools is simply due to the inherent complexity of networked storage.
The category “storage management” is a very board umbrella, encompassing everything from interoperability of products to performance tools and capacity management.
Storage users may feel pain, but the symptoms vary widely from data center to data center. The primary challenge of a board initiative such as SMI is to provide a comprehensive set of management tools within a single framework that can address all fundamental aspects of storage management issues.
Although the SNIA has come under criticism for failing to deliver a viable, all-inclusive solution in a timely fashion, we should not underestimate the challenge its technical volunteers face.
Vendor competition and bickering in the process clearly does not help, but the struggle to make SMI a reality has far more to do with the nature of storage networking technology than the nature of vendors in a market economy.
Just have a casual read through the SMI 1.0 standard or the current 1.2 drafts, and you immediately appreciate the immense effort required to bring storage networking under control.
The formulation of a comprehensive, multi-vendor management standard on its own is of little benefit unless it can be put into productive use by customers. The development of Fibre Channel and iSCSI standards, for example, was accompanied at each step by compliance and interoperability testing.
Some vendors were far more rigorous in this process than others, and a few unfortunately developed reputations as habitual saboteurs. The net result, however, is that the Fibre Channel Industry Association (FCIA) in conjunction with ANSI and the SNIA IP Storage Forum in conjunction with the IETF were able to validate working standards in working products and thus promote a much higher degree of interoperability for users.
For SMI, the SNIA Interoperability Committee’s Compliance Test Program (CTP) and interoperability demonstrations at Storage Networking World (SNW) conferences have helped bring comprehensive storage networking management one step closer to working reality.
Although all major storage vendors are implementing SMI in their products, and although storage administrators are aching for a remedy, SMI-enabled solutions have not yet swept the market.
Standardization, compliance testing and carefully tended interoperability demonstrations do not directly translate into customer confidence and willingness to undergo the potential disruption of changing management platforms.
In addition, fundamentally resolving storage management problems in their diverse forms will require additional functionality that is still under construction. Storage virtualization, for example, is an essential foundation technology for higher level storage services such as automation and information lifecycle management in heterogeneous networks.
Although SMI 1.1 includes basic storage pooling APIs and SMI 1.2 addresses basic ILM functionality, these technologies are themselves undergoing rapid development independently of SMI.
In terms of user adoption, the success of SMI depends on its ability to provide a non-disruptive migration path out of the current SAN management dilemma and the ability to continually accommodate new storage technologies as they mature.
Enter Open Source
In addition to the SNIA SMI, yet another group of vendors (also involved in SMI) have proposed an open-source initiative for SAN management call Aperi.
And, although SMI is providing standards-based, open APIs for interfacing with a diversity of SAN and NAS products, each management platform provider would have their own proprietary application with differentiating value-added features. For companies that like to keep their source code sacrosanct, the very idea of open source is a turn toward godless communism.
Open source implies common ownership and the evaporation of profit. Linux companies, for example, only make their margins on support and development contracts, but final code passes to the community at large.
Whether an open source storage management initiative is sustainable depends on how many large vendors are willing to solve SAN management issues by any means necessary, and how quickly the threat of open source accelerates the work on SMI.
Comprehensive, single-pane-of-glass management of storage networks and the ability to manage the complexities of storage transport and storage placement will, in the end, greatly simplify the lives of storage administrators. It will not, however, remove other obstacles that create rocks and hard places on a daily basis.
Simplified management of a complex, heterogeneous SAN is a great thing, but customers must also rely on support from their major providers.
SMI aside, a particular vendor may elect not to support a configuration that includes, in the vendor’s view, unqualified products. SMI will not help customers who are refused support because their current microcode levels are considered out of date, even though everything else is working fine.
Nor will SMI nor Aperi will stop end users from voicing complaints that actually have nothing to do with the SAN and everything to do with the application they’re running. But a comprehensive SAN management platform will at least eliminate much of the day-to-day drudgery of storage administration and provide a rich foundation for ensuring stable and reliable storage operations.
Tom Clark is director of Solutions and Technologies for McDATA Corp., a provider of netoworked storage solutions.