Sex, drugs, pain and storage

New storage technology can be frankly pornographic: it's big, it's sexy and you want it slammed into your rack right now — but is a long term relationship more satisfying?

New storage technology can be frankly pornographic: it's big, it's sexy and you want it slammed into your rack right now. But just as a long-term relationship can ultimately be more satisfying than a one-night stand, sensible storage management requires a fair degree of planning and patience if you want to shoot multiple goals.

I was reminded of this at the recent Gartner emerging trends roadshow, where analyst David Cearley discussed disruptive technologies and how to work around them.

Part of the reason we don't get into long-term planning, Cearley argued, is because continuous improvement means we simply haven't had to bother.

"Moore's Law is the crack cocaine of IT," he said. "We can deal with increasing data loads because stuff just gets faster."

That scenario is ending, Cearley suggested, because the shift to multi-core processors hasn't yet been reflected in software specially written to take advantage of them: "There are some written with a certain amount of multi-core capability, but not many in enterprise applications."

Storage generally falls into this area — while we may be very concerned with data throughput at a hardware level, when was the last time you benchmarked the basic software operations of your system? And does it contain any multi-core code? I suspect the answer to these would be "never" and "no", phrases which seem oddly resonant with the generally down-market tone of this week's column.

So let's get back to the complications. "The result is you can go from one core to 16 cores and those applications don't get that automatic power boost. You may even see a decline."

The bottom line? Powerful systems no-one can properly use, and a lingering sense of dissatisfaction that we're not getting everything we could be.

The solution is rather less sexy, but ultimately less painful: planning.

"Start doing some inventory to identify which areas will have potential problems," Cearley advised. "Today, you're not going to be feeling a lot of pain. In four years, there'll be a lot of pain. It's not that in 2008 you need to spend a lot of time rewriting everything, but you need to spend some time planning for the long term."