When you design performance tests, your goal is to develop
few use cases that adequately test the functions that are most frequently
used and most important.
Procedure
- Determine how and when users log in and log off in the
production environment. For example, do users stay
logged in all day? Or do users log in, complete some transactions,
and then log off?
- Identify workloads that reflect the workload distribution
and the workload rate. Distribution is the percentage of
users who are completing a specific task, such as creating service
requests. The rate is the transaction rate for a specific task, such
as the number of service requests. Develop use cases that reflect
both the number of transactions and the frequency of transactions.
For example, 20% of users are creating 10 service requests in an hour.
- Determine how user load changes throughout the course of
the day. If you have real-time usage history, use the
data for requests processed per second during peak hours to create
your use case. Then, you can compare your real-time data with the
server load created during the performance test to see how your deployment
is functioning.
- Based on user behavior, workloads, and user loads, create
use cases that focus on your test objectives.