I'm tasked with testing out ProxySQL on a couple of Tomcat test servers, comparing its performance and resource utilization with the current Apache DBCP connection pooling. I've got a rough load testing plan in place, but I have near zero experience with monitoring and gathering/comparing system-level performance metrics (CPU, memory, disk, network, etc.)
My load testing will involve an Ansible playbook that does various preparatory tasks, triggers the load testing scripts, and finally cleans up after itself. I'd like to add "starts and stops the gathering of performance metrics before and after the load test, respectively", but I'm at a loss as to what tools I should look into.
The optimal solution, in my view, would look something like this:
- Runs locally in the background on the Tomcat boxes
- Simple configuration
- Monitoring sessions can be started and stopped on demand
- High resolution (metrics gathered once per second or even more frequently, if need be)
- Results of monitoring runs easily readable into some kind of graphical tool for comparing different load test runs.
Given these specs, what kind of options am I looking at, both for the monitoring task itself and the graphical tool with which to examine them?
Best Answer