Designing an Appropriate Performance Test Environment. Part I

Designing an Appropriate Performance Test Environment. Part I
February 07 10:00 2012 Print This Article

Before you can perform an effectual performance testing strategy it is necessary to take into account a lot of factors. Surely, there’s a lot more to performance testing than simply generating a load and observing what happens.

Designing an Appropriate Performance Test Environment. Part I

But one of the most significant factors is to consider your performance test environment. In an ideal condition, it would be an accurate copy of the deployment environment, but this is very seldom situation.

There are some reasons for this:

  • The number and specification of servers

It is usually unfeasible to provide an accurate replica of the server content and architecture in the deployment environment. It happens so because of the high price and intricacy.

However, even if it is impossible to replicate the numbers of physical servers at each application tier then you can try to match the specification of the live servers. This will permit you to define the capacity of an individual server and provide a reliable baseline for horizontal scaling.

  • Measurement of application databases

The size of the test environment database should be similar the live one; failing which, the dissimilarity will have a significant influence on the validity of performance test results.

Performing tests against a 1 GB database when the live deployment will be 50 GB is absolutely unreal.

  • Number of application tiers

It is often happens so that the number of application tiers has much bigger influence on the validity of performance testing than the number of servers. Consequently, it is importunate to hold the live tier deployment model till there is no other choice.

  • Bandwidth and connectivity of network infrastructure

In a parallel vein, it is rare for the target servers to be deployed in the same location as their live counterparts, although it is frequently possible to share the same network infrastructure.

Consequently, the normal performance test environment is a subset of the deployment environment. Sufficient performance here signifies that things may only get improve as we transfer to full-blown deployment. But unluckily, that is not true in every case.

Related Posts:

  • No Related Posts

About Article Author

view more articles
Nataliia Vasylyna

View More Articles