Why your IT infrastructure’s utilization doesn’t matter (Part 1)

Barry Cooks, Vice President of Engineering –

This installment is part one of a two-part series on the shortcomings and benefits of different metrics for IT infrastructures. In this post, I’ll look at the problems that come with only measuring utilization, and in part two, I’ll cover the greater value that comes with measuring performance.

IT is data-driven, or at least that’s what today’s industry leaders want you to believe. There’s data-driven marketing, data-driven cybersecurity and data-driven campaigning. But what if all the individuals making these data-driven decisions are actually looking at the wrong data entirely?

This is a current challenge in the IT world, due to an all-too-common misunderstanding of utilization versus performance.

Utilization in IT infrastructure is the measure of how well the system is being put to use from a capacity and bandwidth standpoint. For example, when you’re considering storage utilization, you’ll need to look at available storage space, time required to store and retrieve files, speed of data transmission and plenty other variables. This is all useful information for certain operations, but it’s actually not that valuable when it comes to your end users. They don’t care what percentage utilization your system is at during peak demand hours; they just care about whether or not you’re meeting your performance-based service level agreement (SLA).

What do your customers see that you’re missing?

Have you ever been on the phone with a customer, listening to him come down on you about an unavailable application, but you’re looking at your utilization statistics and can’t find any problems? That’s because your utilization numbers don’t truly reflect your system’s performance. And there’s the difference: ‘good’ utilization is an indicator of how well you’ve optimized your purchasing cycles to most efficiently spend your money, but performance is the measure of how well you’re providing a service to your end user. You see, there’s a phrase that goes, “There are lies, damn lies and statistics.” All the utilization numbers and percentages in the world don’t matter if your customers aren’t satisfied, but those numbers are usually where IT teams stop their investigations.

To get a look at your infrastructure’s utilization, you need to take a sample of its operations. But given how fast systems are operating today, even a five-minute summary in a 16 GB network will cover millions of transactions. Calculating the average of that activity isn’t actually giving you a clear picture of the ups and downs going on inside. This means while your customer is complaining about the application downtime, you’re just looking at the positive average utilization numbers, rather than seeing the individual hindered transactions the next level down that are causing real problems. Data-driven infrastructure decision-making becomes less helpful when your information isn’t reflecting the real issues at hand.

Interested in learning what your company can do to solve this problem? Follow us on Twitter at @Virtual_Inst and be on the lookout for part two of this blog.