Process Improvement at Italtel

Italtel Linea UT has been conducting a software process improvement programme since 1991, using approaches like: process certification in accordance with ISO 9001, Process Quality Management and Improvement (PQMI), process assessment in accordance with Bootstrap. After four years, improvements were tangible and it was possible to set and track quantitative goals for the projects.

Background

Italtel designs, manufactures, markets and installs systems and equipment for private and public networks, including the Linea UT telecommunication switching system. At the end of 1993, about 15 million Linea UT lines, 1000 exchanges and 2000 RSU (Remote Subscriber Units) had been installed in 18 countries world-wide. In order to define priorities for the improvement programme, in November 1991 a process assessment activity took place at Italtel Linea UT. The assessment adopted the Bootstrap approach.

The Assessment Results

The assessment team concluded that, as of November 1991, Italtel was operating in a controlled way, corresponding to a maturity level at that date of 2.5 (on a scale ranging from 1 to 5) both at SPU (software producing unit) level and at project level. In order to boost process improvement and thus evolve towards a stable maturity level 3, the following recommendations were made:

the impact of quality assurance practices at project level ought to be strengthened, together with the creation of a Software Engineering Process Group (SEPG);
review of the Quality Management System should be undertaken, in accordance with ISO 9000 standards;
introduction of product quality planning and tracking (with validation of non-functional requirements using the ISO 9126 model);
fine-tuning of standard operating procedures;
strengthening of software testing practices, including the set-up of automatic non-regression suites;
extension and broad application of the measurement system.

The Improvement Programme

The overhead of the improvement programme accounted for about 5% of the overall man-power of the Software Laboratory, including SEPG staff, ad-hoc groups, work on standard operating procedures, data collection and analysis, and training of all staff.

One of the key assets of the improvement programme was the metrics initiative. The main goals of the metrics programme were:

to provide the quantitative information essential to identify opportunities for improvements;

to verify the advantages of the implemented changes to the development process.

The metrics were defined in accordance with the goals of the SPU: timeliness, reliability, documentation, resources. Indicators were collected and analysed from two different points of view: the customer and the project.

Results

Considering each of the goals, the following improvements were observed, both analysing differences between 'before' and 'after' metrics and calculating the net benefits from process improvement:

Timeliness

timeliness for the customer had always been achieved; this must be considered an excellent performance by the SPU; if the net contribution of software improvement is considered, it can be shown that timeliness capability had been improved by a factor of 10%.

Reliability

fault density in operation has dropped to less than 1 fault per 10,000 LOC, with an improvement of 54%, and an increase in the "test effectiveness" indicator of 7%. Positive results come mainly from the improvement action of automating the non-regression test suites. As stated in a number of published studies, a fault found in the field costs 10 times more to correct than one found in testing; using this assumption, the improvement in test effectiveness brings a saving of 27% in the effort for corrective maintenance. The net contribution to process improvement shows that in this area, capabilities have increased by 20%.

Documentation

the availability of documentation both for the customer and for the project has quickly increased and has reached its goals, without negatively affecting other indicators (in particular: timeliness issues);

Productivity

goals have been largely reached and we can observe better performance by a factor of about 65%; nevertheless, if we consider the difference between expected values and actual ones we can conclude that this is the area where the benefits from process improvement are less tangible.

Conclusions: a Bootstrap Re-assessment

In May 1994, 30 months after the first assessment, a second Bootstrap assessment took place in order to monitor the effects of the improvement plan. The findings from the re-assessment indicated that substantial improvements had been obtained. Italtel Linea UT had progressed to a maturity level 3.5 at SPU level and 3 at project level, with many activities reaching a level 4 grading. Improvements occurred in all areas, indicating that the SPU had reached an excellent position in international terms (according to the latest SEI report on the subject, only 7% of assessed companies can exhibit level 3) and in particular in Europe (in comparison with the ratings stored in the centralised Bootstrap data base of assessments).

Assessment results at SPU level: 1991 versus 1994

References

Damele. G., G. Bazzana, j. Rozum, J. Siegel and D. Zubrov, Quantifying the benefits of software process improvement in Italtal Linea UT exchange, in: Proceedings of ISS 95 - XV International Switching Symposium, Berlin, April 23-28, 1995