Developing a DSL-based Load Generator for Timeseries Processing Systems

Performance testing of time series processing systems requires a high amount of technical
knowledge. Some stakeholders in the software development process lack the required
knowledge and therefor cannot understand let alone write performance tests for their
systems. Also, writing performance tests in programming languages is time consuming.
To target those problems, this thesis proposes a domain specific language that focuses
on usage even without much technical knowledge. It is also designed for ease of use to
reduce the necessary time to write a test.
To make usage of our domain specific language possible without much technical knowledge,
the stakeholders were divided into two groups. Developers with much technical
knowledge do a onetime configuration of the systems used in the tests. Users can then
write performance tests for the configured systems using our domain specific language.
An interpreter for the language realizes the expected behavior. To ensure future changes
and additions can be integrated into our solution, it is designed to be highly extensible.
A small user-study was performed to evaluate the usage as well as advantages and
disadvantages of the proposed solution. The results confirmed the desired ease of use of
our language.