IT, automate thyself
December 04, 2006
One of the big trends in IT management is the drive to improve employee productivity in the data center, where labor costs can run as high as 50% of total costs. The new edition of the Financial Times Digital Business Podcast, out today, includes a commentary by me on how, as the FT describes it, "automation is about to rip through the IT department." The podcast also features the FT's Lucy Kellaway offering a timely defense of shopping online while at work. You can listen or subscribe here.
Nick, it's a very nice idea that IT shops will spend a lot of time and effort in reducing labor costs over the next year — at Forrester we regularly find that an average US IT budget is nearly 40% staffing — but do you think this is really possible? While a firm like HP can certainly save a billion dollars consolidating data centers, my educated guess is that firms with this sort of major infrastructure are few and far between. While I do believe advances like SOA can reduce the demand for labor, you (of all people!) should know that the cost of labor is equal parts demand and supply.
On the labor supply front things are pretty ugly. UCLA's Higher Education Research Institute has put interest in computer science among incoming freshmen at it's lowest level since 1977! New majors entering the field has fallen 70% since its peak in 1999. Unless we can come up with some serious breakthroughs in IT automation, or find some untapped qualified labor pool, I think we're looking at a tightening high tech labor market, regardless of HPs data center consolidation. By my calculations that will equal higher costs, not lower.
Good points. You're right that if supply tightens then labor factor costs will go up (ie, an hour of a worker's time will cost you more), but that just makes improving labor productivity (ie, getting more work done for every hour of worker time) all the more important. If you have to pay more for a worker, in other words, then you want to have fewer of them.
I do think we're at the point that extensive automation of traditional data-center work is becoming possible - and will happen. Big companies (and, importantly, big IT utilities and outsourcers) will be the first to reap the benefits, but the benefits will flow to smaller firms in time.
Whenever I hear these type of prognostications, I have to laugh. IMHO, it almost never the technology (or the technologists) who are the obstacle to automation. Normally, the biggest problem is that the organization as a whole can't be specific enough about what they want or aren't willing to pay for it. Also, once everything is automated, the people who insisted on it are often the people who complain the most when they discover that takes time to make changes in the future.
Unfortunately, the people who listen to FT's podcast will now be pushing the CIO to get some of that thar 'au-tow-ma-shun'. Instead of focusing on the CIO and cost per hour, we should be focusing on the long-term success of the organization, in terms of flexibility and profitability.
I'm always happy to hear this topic come up. I'm the author of Puppet, which is an open-source server automation framework, and I'm currently at LISA, which is USENIX's yearly sysadmin conference, where we've so far spent the week talking about automation tools and practice.
I think the main thing driving automation right now is that companies just aren't able to deliver required services without it. Just like you can't afford to build your own operating system any more, you can no longer afford to build your own infrastructure from scratch. Those who do suffer from very low service quality.
Unfortunately, this is still basically the standard -- very few companies have any automation to speak of, and those that do have pretty simplistic versions. It's surprisingly difficult to get anyone to even talk about the problem, much less invest money in trying to solve it. Notice, for instance, how little investment there is by VCs in this space.
So, one of my questions would be, what tools are going to be used to fuel this massive spread of automation? Because they don't exist right now, and I don't see many people working hard at creating them.
Posted by: Luke Kanies at December 5, 2006 10:21 AM
Post a comment
Thanks for signing in, . Now you can comment. (sign out)(If you haven't left a comment here before, you may need to be approved by the site owner before your comment will appear. Until then, it won't appear on the entry. Thanks for waiting.)
"Riveting" -San Francisco Chronicle
"Rewarding" -Financial Times
"Ominously prescient" -Kirkus Reviews
"Riveting stuff" -New York Post