Other Aspects of Performance Management

Problem Management

Leaking pipe

Capacity ManagementFish in a small bowl

Performance EngineeringSpeedometer






Service Level Management

Exasperated userLife can be frustrating at times. Now that computers are part of our everyday life, they can add to that frustration. Anger at your computer screen is a new type of road rage.

No one wants their users or employees to have this type of frustration. Avoiding it is what service level management is all about.

What is Good Service?

You might think that you could look at performance data from a computer and determine if it is providing good service. But you would be wrong. Processor and I/O utilizations, for example, might tell you about the load on the system, but they don't tell you if it is providing good service. The reason is that "service" is subjective. It depends on your goals.

Is a system that provides a response time of 0.8 seconds 99.9% of the time providing good service? If its a web site, then it probably is. But what if it's a real-time system controlling a nuclear reactor? In that case the response time might be disastrous. It's up to you to specify what is good service.

Service Level Requirements

To manage service levels, you need to start by defining your service level requirements. With well defined service levels, you can always tell how your system is doing. At least you can if you are measuring the service levels.

Defining service levels is both an art and a science. They need to capture what is acceptable to the users of your system, as well as any governmental or organizational requirements.

Workloads and Service Levels

Modern computers run many systems. It is rare that a computer system has only a single service level. To deal with this reality, the applications running on the system are broken into "workloads," each with its own service level requirement.

Ideally, service levels would be defined when the requirements for a system are developed. In designing and building the system, performance engineering would be used to make sure the system meets its service levels. In practice, this isn't usually the case.

Having workloads with service levels is also required for capacity management.