Performance categories and characteristics
What is a performance category?
A performance category is a group of related system performance characteristics to provide a more focused understanding of a group of system behavior. For instance acceleration and top speed of a car is a category of performance characteristics. Transferred on system performance of an IT system this could be processing capacity and and responsiveness, which are some of the characteristics in the category powerfulness.
Each performance category contains several performance characteristics and each characteristic can be represented by one or more performance metrics.
Every system performance metric belongs to one or more of the following three system performance categories:
Powerfulness are measurements of a system’s delivery limits, i.e. how much services can be produced or how fast can services be delivered (on the tested platform).
Powerfulness characteristics are usually the only association made to a system's performance. It's like limiting a car's performance characteristics to top speed and acceleration, without mentioning other important performance characteristics, such as fuel consumption per 100 kilometers, service interval etc. To get a full picture of a system's behaviour additional performance categories must, therefore be studied.
Reliability are measurements of a system’s conditional limits, i.e. the circumstances for maintainable service levels.
Reliability characteristics are measurements of a system's behaviour over time and during critical circumstances, such as resource shortages, overload situations, or partial or full restart. Figures should provide a full picture of a tested system's expected long term behaviour.
In the connected world, reliability characteristics are rapidly getting more and more important. This is a simple consequence of more systems are providing its services on a 24/7 basis, i.e. the services shall be available globally at any point in time.
Why are reliability related characteristics complicated to measure?
Reliability related characteristics of a system takes a long time to measure and require complex test case definitions with masses of predefined data to customize service requests by simulated users. Like all system performance measurements, tests will block the test equipment from other usage during during the test execution. In the case of reliability measurements the execution time can be really long (weeks) and must therefore be planned carefully.
The reward is a rock-solid system that can handle system critical situations that would otherwise put the system out of business.
Why are reliability related problems hard to identify and correct?
A reliability related misbehaviour of a system is unfortunately hard to reproduce and therefore hard identify and time consuming to fix.
Consequently unreliable systems usually face a short life cycle in the connected world.
Efficiency are measurements of a system’s productivity limits, i.e. required resources for different levels of service production.
Efficiency characteristics are measurements of a system's resource usage for producing different services. Efficiency characteristics are measurements of a system's resource usage for producing different services. This can easily be compared with the fuel consumption per 100 kilometers for a car. The less fuel consumption the more efficient characteristics.
In the connected world where reliability characteristics are rapidly getting more and more important, work on a system's efficiency in service production is the key to a robust and reliable system.
Efficiency characteristics are usually easy to measure and shall be done in combination with powerfulness measurements.