Chapter 3. Database Hardware Benchmarking
After all the theory in the last chapter about what makes some systems perform well or poorly, you might be wondering just how your own system measures. There are several reasons to always do your own hardware benchmarks. The first is simply to be systematic about your performance process. If you always measure things yourself, you'll get a better feel for what good and bad performance looks like, one that can help tip you off to even subtle problems.
Second, in the case of disks in particular, problems here are a very common, underlying cause of database performance issues. If your disks are slow, and there are many ways that can happen, your database will likely be slow too. It's important when this happens to have accurate data on whether the problem is likely at the hardware or software level.
The goal of your basic hardware testing should be to look for large configuration errors, not to analyze every possible parameter...
