Measuring performance

When we start to have a remarkable number of resources on a node (in the order of several hundreds or thousands), compilation and application time of a node catalog grows to uncomfortable levels.

If the number of the nodes to manage is big, even small tunings and optimizations of our code can bring interesting results.

For this reason, it is useful to have tools and techniques that permit us to measure Puppet's performance metrics at our disposal.

Puppet Metrics

Puppet itself provides some options that help us understand where time is spent during its activities.

At the end of each Puppet run, it is possible to see a detailed report on the time spent for each kind of activity; on puppet.conf, we can enable reports with the following option:

report = true # Enable client's reporting

We can have a summary of the run times with the following option:

summarize = true # Print a summary of the Puppet transaction

At the end of a Puppet run, we can have metrics that let us understand...