WATCHING THE WATERS

Enter the mysterious, and much neglected, world of performance monitoring. The mystery and neglect have the same root cause; performance analysis is a very complex issue that can be quite daunting. Performance engineers typically deliver thick reports replete with statistical equations, graphics, and polysyllabic run-on sentences to explain why you will need to replace your current infrastructure within a given time frame. System administrators yawn at the lengthy explanations and statistics equations that, frankly, can look like recipes written in Martian. You may be wondering why you really need to worry about performance monitoring for Linux since hardware is at almost “throw-away” prices today. The answer is that, for larger installations of Linux and mission-critical applications, it is less expensive to add to your infrastructure than to replace it. With clustering, fail-over applications, and distributed environments supporting hundreds or thousands of users, adding another 4 CPU server with 4GB of RAM is infinitely more palatable to those who write the checks than replacing everything you have because you have run out of resources. By collecting performance data, you can make informed decisions about upgrading, adding to, or replacing hardware as business growth dictates.