Unisys flags up customer storage waste

It Might Be Nice To Know for Curiosity's Sake But...

Underutilization is only one way to describe it. Other ways are reserve capacity, hot spares, distributed risk, distributed load, and many other good things. Its not a good thing to run at 95% utilization. And if your controllers are smart enough to power down drives after a period of disuse there is minimal power loss and a corresponding extension of life.

I know its not exactly the same thing but I'm actually trying to move more toward a lot of lightly used hard drives in my servers. The drives get more breaks and are less prone to overheating, there is less impact on performance from multiple users accessing different data sets, etc. Peformance is awesome when disk intensive programs actually read from and write to two entirely different controllers. Two green 5400 drives can outperform one 10K drive for typical server tasks and they cost less, provide higher capacity and generate less heat. All the drives last longer because they spend a lot less time seeking and thrashing.

I always laugh when vendors try to force you to put all your eggs in their one overpriced basket when you can buy 10 commodity baskets cheaper and get a lot of side benefits. Their main selling point is usually "reduced management costs" but this study is pointing out that the management component is failing. For them to achieve a savings of 50% on power implies that you have probably 3x as much storage as you need, it shouldn't be too hard to find the space in that case. If you only have 2 times what you need you won't be able to turn off half of it when you consider controllers, switches and such.