NMS product line is responsible for developing a Cellular Network Management System NMS/2000, which is a large-scale system consisting of almost two million lines of C++ code. The system is developed and maintained incrementally in subsequent projects so that the new system is built on top of the existing one. As the need to create customer specific product variants increases in the near future, the demand for testing is rapidly increasing.
We see that the only alternative to survive with satisfactory development lead times is to improve test management, execution and analysis. Also motivational issues in testing large systems on teams can be better handled when most repetitive testing actions are automated and human creativity can be used for more demanding and interesting tasks, such as test design and fault analysis.
The objective of the Process Improvement Experiment (PIE) is to decrease lead-time and to increase testing efficiency in relation to product complexity and size.
In order to meet these objectives, we take separate aspects of testing automation tool support into use in all our four defined testing phases. That includes four separate but partly parallel tasks:
1. a) Collecting the four testing phases into a coherent testing process that sets the requirements for test automation tools and on the other hand b) making the road map for necessary process changes to get any real value of the tool in the longer run: increase testing repeatability and test case traceability, define configuration management policy for test scripts, etc.
2. Evaluating and selecting tool for test automation
3. Building necessary process support for creating tests with the tool: training, tool usage guidelines, testing script libraries, etc.
4. Piloting the tool in parts of two baseline projects, T9 and T10 (around 40 person-months each)
The achievement of the objectives is assessed in testing process assessment before and after the PIE and also in audits during the pilot phase.
The total personnel of 250 in the NMS product line includes 140 software engineers in the baseline projects. The great success factor in favour of the PIE is our PI mind-set: persistence of the process improvement effort with management leadership, dedicated quality and process improvement groups and usage of PI framework.
EXPECTED IMPACT AND EXPERIENCE
We do not expect the piloting of the selected tool to make big quantitative differences in the productivity or efficiency of testing in the PIE time frame, because biggest bottlenecks seem not to be solvable through test automation during the PIE. We hope for bigger gains in the process side: after the PIE we have gained more repeatability in the testing and understand better the potential and demands of test automation in our context. Also as a side-effect we will have better analysed non-technical bottlenecks in the testing process.