System Performance Monitoring with SE
A basic reason why we monitor the performance of a system is to define the overall state of the machine (“it’s OK” or “it’s slow,” etc.). This monitoring breaks down into the states of the main components of a system such as disks, nets, RAM, and so on. I have produced a set of performance rules that you can use to find out what state you are in. These were originally described in Appendix A of the first edition of this book, but it was hard to obtain the data and follow the rules by hand. I have coded all the rules in SE and no longer attempt to describe them the way I did before. The SE rules code defines the rules unambiguously, and you can test them by running monitoring scripts that use the rules.