Determing the value of investing in SRM softwareBy Mike Karp
I have a heavy prejudice in favor of storage resource management in general, and standards-based (which is to say, SMI-S) SRM in particular. If you can't discover it, you can't monitor it, and if you can't monitor what's out there… well, good luck with the management.
As is the case with all tools that contribute to automating IT, the value of an investment in SRM software increases with the complexity of the systems the software helps to manage. Highly complex environments benefit greatly from automation; less complex environments extract proportionately less benefit.
It doesn't take deep insight to understand why it works this way. Consider such fundamental "gotta haves" as:
• Asset discovery
• Asset management
• Capacity management
• Configuration management
• Events management and alerts
• Performance management
• Policy management
• Quota management
• Removable media management
Every time you automate one of these, you decrease the chance of operator-induced error. Additionally, things happen more quickly, more accurately and are more repeatable.
Complex IT situations such as grids, on-demand or utility computing - in fact, any kind of dynamic system - invariably win when automation is introduced.
A problem lies within one of my earlier statements, however: "less complex environments extract proportionately less benefit." While I'm convinced this is true, no one can tell us what that proportion might be. I have my own suspicions, of course, as to what the "value decay" might be for an SRM investment. I think the value scales or decays more-or-less in accordance with what our networking brethren know as Metcalfe's Law: "The power of the network increases exponentially by the number of computers connected to it."
Replace "power of the network" with "value of SRM" and you may have a useful metric for defining how much value SRM will have at your site.
Whether or not you like my little law, you have to admit that it does raise at least one question worthy of note, namely: if value diminishes as environments get smaller, how small does an IT installation have to be before SRM is simply not worth the effort?
The answer to this question has, I think, little to do with mathematics or physics, and a great deal to do with human vision: if you can see all the lights, dials and whirligigs on all your machines from where you sit, SRM is likely to provide poor payback. In other words, if you have only a few servers that are supported by a simple infrastructure, you are probably better off spending your money on almost anything other than SRM. On the other hand, if you can't see the front bezels and indicator lights of all your assets from where you sit, SRM may have something to offer.
At least as far as storage is concerned, this guidance also offers a fairly accurate definition of where to draw the line between the "S" and the "M" in SMB. If the vendors understood this, they would waste less of their time and yours with misdirected marketing campaigns.