SQLServerCentral Editorial

Knowing What's Normal

,

I noticed in Database Weekly this week that I found a lot of posts on performance related topics, especially hardware ones. There was a note about a new Intel chip, getting performance counters with Powershell, and a review of drive performance by one individual. There was even some research from Andy Warren on SSD drives and a pre-sales evaluation of whether they are worth the cost. Those are all good topics, and worth reading about. However knowing all those tricks and techniques might not help you unless you're Gregory House, DBA.

There was a great post from our resident SQL Server MCM, Robert Davis, on taking a medicinal approach to troubleshooting performance. Using the television show House as an example, Robert gives reasons why you don't want to necessarily take the approach of just trying something, and that you ought to be aware if you really have a performance problem.

What is "slow" performance? How do you know things are slow? Slow is one of those descriptive terms that's vague, unreliable, and uncertain in it's meaning. I run every day, but there are times that my two miles seem to go faster than others, and times that it seems to go slower. However often I find that my perceptions aren't correct and the times are very similar. I just can't tell what "slow" by myself.

Neither can users. They often don't know even what normal is. Their perceptions, like mine, are clouded by many external factors. Without actually measuring what a normal baseline is, they can't necessarily be sure that their systems are operating any better or worse than on another day.

I'd urge you to not only read Robert's post, but also make sure that you are capturing baselines on your systems, especially the hosts for virtual machines. That way you'll be able to make some objective comparisons and analysis to find out if things really are slow.

Rate

You rated this post out of 5. Change rating

Share

Share

Rate

You rated this post out of 5. Change rating