The Storage Management Initiative Specification (SMI-S) is slowly making its way through standards committees and is starting to show up in vendors' products. Although SMI-S is laudable in every respect, I think it's important for users to take a sober view of the standard and when they'll be able to reap the benefits.
In the near term, it seems that SMI-S will be of most value to vendors—not end users—because it will improve time-to-market and make it easier for software vendors to develop code for managing multi-vendor storage environments. For users, SMI-S will eventually enable more freedom of choice in selecting management software and improve interoperability. The standard may also lead to lower prices because of increased competition.
However, SMI-S faces several hurdles. For example, there are a number of different ways for vendors to distribute SMI-S "providers" (which act as "translators" between hardware and software). This is a topic we'll explore in next month's issue.
Another hurdle relates to legacy hardware. Does anybody really expect vendors to retrofit the installed base of legacy hardware for SMI-S compliance? And if the installed based isn't SMI-S-compliant, then the standard can only be used in environments with recently purchased hardware. The same goes for installed software, because SMI-S is a two-way street that requires compliance on both the hardware and software side. As such, it could be many years before SMI-S-based management of large heterogeneous environments is a practical reality.
Another sticking point is that vendors with the largest installed bases and proprietary lock-ins are generally loath to adopt something that will "level the playing field," which SMI-S does, to a degree. Do you really think that vendor A is keen on making it easier for vendor B to manage vendor A's hardware? I don't think so.
Another hurdle: As with any standard, SMI-S only provides a relatively base level of functionality and vendors can adhere to the standard while adding many levels of functionality (extensions to the standard) that may lead to interoperability problems. We saw this in the early days of SCSI, Fibre Channel, etc.
As a result of these and other factors, some industry analysts don't think that "robust" SMI-S implementations will appear for another two to three years. Of course, it's impossible to predict when any standard will be ready for prime time, but in the storage industry it's usually safer to bet on the long-term view. Consider iSCSI: The spec was finished in 2000, and we're only now seeing a trickle of end-user adoption.
So, by all means pressure your vendors to implement SMI-S, if only for the check-off item in the RFP. Eventually you'll realize some benefits from the standard, but don't expect SMI-S to solve your management headaches overnight.
If you're interested in SMI-S, consider attending the Storage Networking World conference and exhibit next month (Oct. 25 to 28 in Orlando, FL), which will feature multi-vendor SMI-S demos and a variety of presentations on the standard. For more info, go to www.snwusa.com/nww.