Four ways to cut datacenter costs

[ Bill Snyder is on vacation. David Strom is substituting until Bill’s return. ]

With the markets in a tizzy yet again this week, I have my own calming strategy for you: Start thinking about ways you can cut the cost of operating your datacenter. Here are four suggestions — if you can tear yourself away from watching your stock portfolio bounce up and down.

Think modular. Instead of buying those sexy shipping containers that have their own cooling and pre-mounted systems, consider how you can make do with a similar arrangement on your own raised floor by reducing the air-handling and cooling requirements. The more modular you can make your own datacenter, the lower your monthly energy bills will be. Granted, you might not have much flexibility in rearranging your equipment racks, and that raised floor might be so stuffed with aging cabling that there really isn’t much room for any air to flow to your equipment. But that is all the more reason to start thinking of how you can consolidate a few racks into a more modular chunk.

[ Looking for other ways to trim? Then have a look at Tom Sullivan’s “Five outside-the-box ways to cut IT costs.” ]

Eliminate or reduce server service contracts by having warm spares that are ready to go. This DIY strategy works only if you can buy similar equipment in quantity that really does have the same innards. The trouble here is that it’s tempting to just put the spare into production and not replace it immediately. If you take this strategy to its ultimate conclusion, you end up managing your peak computing demands by creating virtual servers. When you need a spare, you use one of the numerous dynamic provisioning systems, such as VMware’s Virtual Lab Manager or Novell’s Platespin to bring up a new virtual instance of your server. A new company, Skytap, is out with its own solution called Virtual Lab Platform that supports a number of virtualized environments including Citrix Xen and VMware and also has a set of programming interfaces so you can integrate your own resources with virtual ones.

Use a managed services provider to bring all of your servers up to the same OS and patch level so you are free to do other things. While you might think this adds cost, it has real long-term benefits, as the customers of companies like Clearpoint in Little Rock, Ark., have found out. The savings come in reduced downtime and staff overtime, as well as recapturing all those hidden moments when your staff are trying to bring a dead machine back to life because of some driver conflict or patch that was misapplied. For example, your provider’s staff can wait on hold to get an answer from a vendor so your people don’t have to. This gets you out of the plumbing business and gives your people more time to devote to actually servicing mission-critical applications. If you haven’t thought about using an MSP for anything yet, here is a great place to start that offers basic questions to ask any potential provider.

Exercise remote control. Finally, get better at using the latest in lights-out remote management tools. You may already think that you have this under control, but there are always new ways to remotely manage the equipment you have, without a lot of additional cost. The better you get at this, the less time your maintenance staff will have in traveling back to your offices in off-hours to fix something.

Let me know what other suggestions you might have for trimming ops costs. And by the way, did you see what happened with GM this week?