Taneja Group Research: The Costs of Performance
What's an IO worth to you? Is it worth more than a gigabyte? Less? That's a hard question for many IT and business professionals to begin to consider; yet we often see it bandied about. It certainly has merit, it just isn't easily understood.
The fact is there are few among IT professionals who realize how often IOs play into application performance, nor what they really need in terms of IO behind every application. Such quests are typically left to storage experts, and when it comes to performance, the industry has seldom equipped the storage expert with much better than rules of thumb, current performance figures from systems that may be riddled with bottlenecks, and magical guesstimates.
Yet the relationship between applications and storage performance is the critical foundation underlying the usefulness of applications ranging from in-the-microsecond decision support systems to time-consuming computational design systems. Behind such applications, minor IO latency or throughput differences can turn cascading reactions into wasted time day-in and day-out, missed revenue, missed goals, blown schedules, competitive inadequacy, lost customers or any other number of seriously bad possibilities.
For the first time in decades, we're seeing good hope for better storage solutions, and the impact on businesses stands to be enormous. That's because there is a huge cost hidden in how today's storage technologies are used to deliver performance, and better technologies will dramatically alter the costs of many parts of the storage infrastructure. That makes understanding these costs, and how they might be altered, more important than ever. We'll take a look at how big the cost of performance is, and with that understanding in mind, we'll look at two examples of new solutions. What they suggest is a new way to get cost-effective performance inside the data center walls.