Recommended Query/Load Analysis Stack

Hi All,

I’m looking for recommendations on what everyone is using for query/load analysis on MySQL (or Percona Server, of course).

We are already aware of (and use) pt-query-digest, and we also use NewRelic. Both of these tools have quirks, but the main problem that we’re having is that currently, separating the signal from the noise is getting to be very challenging, particularly under high load. We’ve fixed the worst outliers that both of these products show to us, so what remains are a large segment of nearly-the-same lesser offenders (in terms of slow queries). This can make finding the root cause of any ongoing problems quite difficult, even when using pt-query-digest over a very restricted time-period (i.e., the last fifteen minutes). Generally speaking, as everything slows down pt-query-digest reports more and more noise, submerging the “signal” even more.

I’m wondering if there are more advanced toolsets out there which are better at root-cause diagnostics and analysis for high query load situations, and if there are, what they are.

What I’d be looking for is an up-to-the-minute display of database performance, and most especially query flow and load metrics that make it easier to figure out which particular query is the root cause of ongoing database performance issues.

Alternatively, if there are more advanced techniques that you are aware of that can be used to more easily separate the signal from the noise in pt-query-digest during high-load situations, I’d love to hear them.

Thanks for any suggestions or recommendations that you can make!