Performance

Engineering
Services

What Is QSEM^SM?

QSEM is a model-based approach to quantitatively evaluating the scalability of Web-based applications and other distributed systems. It was developed by Dr. Lloyd G. Williams and Dr. Connie U. Smith internationally recognized leaders in the field of software performance engineering (SPE). QSEM uses readily-obtained data from straightforward measurements of throughput at different numbers of processors or nodes to match the observed behavior of the application to one of several scalability models. The model can then be used to extrapolate the behavior of the application as more resources (such as CPUs) are added.

Knowledge of the scalability model that best describes an application also provides insight that makes it possible to identify and evaluate software modifications that might enhance the application's scalability.

The QSEM^SM Method

The QSEM method consists of seven steps:

1. Identify critical Use Cases–Identify the externally visible behaviors of the software that are critical to responsiveness or scalability.
2. Select representative scalability scenarios–For each critical Use Case, identify the scenarios that are important to scalability.
3. Determine scalability requirements–Identify precise, quantitative, measurable scalability requirements.
4. Plan measurement studies–Identify the bottleneck resource, plan measurements, develop load generator scripts, determine what parameters to measure, identify needed measurement tools, and document the test plans.
5. Perform measurements–Conduct the measurement experiments, collect data, and document the results.
6. Evaluate data–Evaluate the measurement data to determine whether the scalability requirements can be met and select the best scaling strategy.
7. Present results–Present results and recommendations to stakeholders.

The first three steps are concerned with gathering the information needed to perform the measurement studies and evaluate their results. The next two steps address planning and carrying out the required measurements. The last two steps focus on evaluating the measured data and presenting the results to stakeholders.

Once the results have been presented, stakeholders can use this information to select the scaling strategy that best meets their needs.

Quantitative Evaluation of Scalability – Why?

While scalability is important to today's software applications, few organizations understand how to quantitatively evaluate their software's scalability. As a result, they often make assumptions about the scalability of their software or optimistically assume that it will be linear. If wrong, these assumptions can be costly.