22.3 Experiences in Implementing Configuration Management

This section provides some testimonials from companies that have experimented with introducing configuration management. At the time this book was written, not many controlled experiments existed on configuration management process improvement. However, the European Software Institute's SISSI ( The Business Benefits of Software Best Practice, Case Studies ) project has produced a number of short reports . Of these, five had configuration management as their subject. The contents and conclusions are presented in the following paragraphs.

The European Software Institute's Web site (see bibliography for the URL) contains European best practices for many processes within software development. See also Introducing ClearCase as a Process Improvement Experiment, a Norwegian survey of the profitability of introducing configuration management.

Overall Conclusion

The conclusions of these reports indicate some trends:

  • Introducing configuration management is a great advantage.

  • Management support is essential.

  • Pilot tests of configuration management processes are important before rolling them out on a larger scale.

  • Introducing configuration management is difficult.

Datamat Ingegneria dei Sistemi

Case Study: 10564

Project title: Introduction of Configuration Management

Description: Gaining a competitive edge. By introducing configuration management into the development process of their financial application products, Datamat Ingegneria dei Sistemi S.p.A. vastly decreased the time-to-market and the number of errors in their software products. The overall effect was to decrease development costs so that Datamat could gain a competitive edge.

Lessons Learned:

  • Top-level management support is essential. Introducing formal processes where there used to be none can often generate opposition . Before seeing the benefits of introducing configuration management, it is first necessary to introduce what may seem like useless bureaucracy for a time. Top-level management support is necessary in order to make some team members believe in the importance of the task.

  • Choose a significant project. In order to see the pay-offs from implementation of configuration management, it is advisable to choose a project with at least a moderate degree of complexity where the development team is big enough to benefit from a formal system of collaboration.

  • Take a realistic view of working practices. It is easy to get carried away with the possibilities of a sophisticated software process model, risking failure due to its unworkable complexity. Datamat learned that it is better to start with a basic number of object states and roles, and then elaborate on that basis later if necessary.

  • Buy a configurable product. The wrong configuration management tool can become a straitjacket, forcing your developers to follow a process model designed on the basis of "one size fits all." It is better to invest in a tool with more sophistication than you need (in terms of customizability ) rather than less.

S.I.A. S.p.A.

Case Study: 21244

Project title: Applying GQM to Assess CM Practice for Better Interbank Services

Description: An effective and valuable measurement system is now working. SIA has succeeded in the implementation of the MIDAS project aimed at improving the reliability and availability of interbank services of the National Inter-bank Network of Italy, by establishing an effective Configuration Management (CM) process and defining a suitable measurement program based on the Goal-Question-Metrics (GQM) technique. This has led to higher visibility and ownership of the software maintenance process, supported by precise weekly anomaly reports.

Lessons Learned: Among the most interesting lessons learned in the running of the SPI project, the following can be outlined:

Technological point of view

  • Establishing a CM process is not a trivial process. Therefore it has to be accurately managed; in particular modelling the production process is very critical, especially for companieslike SIAthat were at level 1 of CMM.

  • A considerable amount of time and effort must be dedicated to the selection and customization of CM tools, because there are a great number of complex tools available and rarely do the features provided perfectly fit company needs.

Business point of view

  • Management support is constantly required in order to convince the reluctant ones, mainly when the people responsible for introducing CM are not in a leading position.

  • Constant support after the initial training must be given for CM tools and practices; otherwise people could revert to the old way of working.

  • Incremental deployment of the tool was a winning factor. In this way, only relatively small amounts of software at a time were moved into the CM environment, and only a few people at a time had to be assisted in their initial impact with the new procedures.

Istiservice, S.p.a.

Case Study: 21269

Project title: Configuration and Change Management to Rising Quality of Service

Description: The change management process is now centralized and controlled. The implementation of this PIE [Process Improvement Experiment], focused on the Software Change and the Configuration Management system (SO.C.CO.MA), has led to establishing a centralized Change Management Process that really works and allows the management of environments with a combination of internally and externally developed software. Furthermore the introduction of quantitative measures about the new process represents the starting point toward a management by metrics.

Lessons Learned: The following lessons learned can be stated, among the most significant ones are:

  • The approach can be taken to other processes (e.g., development, testing), by defining a set of proper metrics, establishing a similar collection of procedures, and defining quantitative goals to achieve.

  • If the project were to be repeated, before starting the experiment, the indicators to be taken into account should be more clearly identified.

  • A continuous sensibilization action shall be promoted through specific actions in order to minimize the most frequent risks, such as loss of interest, perception of the project as bureaucratic overhead, and so forth.

  • The new process cannot be considered established; therefore a project review is mostly advisable at the effective start of the application phase.

Event A/S

Case Study: 21379

Project title: Introduction of Configuration Management in Very Small Organizations

Description: Improved customer relations. Through the introduction of configuration management three small Norwegian companiesEvent AS, TSC AS, and Aktuar Systemer ASwere able to increase customer confidence in their products, as well as decrease the time taken to find and rectify errors.

Lessons Learned: There were many lessons learned during this experiment. The main lessons were

  • The process of implementing CM routines is a very complex and time-consuming process. It is an iterative process that requires continuous refinement.

  • CM routines need to be tested in a controlled, but real-life environment before implementing them on a large scale in the whole organization.

  • The identification of appropriate data to quantify possible improvements in the software development process was difficult. Existing information should be used as a basis for defining metrics.

  • An operative CM system has a concrete impact on customer relations. TSC has experienced positive feedback from customers inquiring about their CM routines.

  • Document every stage of the process followed during implementation and ensure that clear procedural documents are produced. This will establish a practical way of operating the system within the company.

  • Make sure that all personnel using the system are fully trained before using the system. This will ensure that once the system goes into operation, all personnel are comfortable with it and know how to operate it. This will ease acceptance of the system.

Sysdeco A/S

Case Study: 21568

Project title: Introduction of a Common Configuration Management Framework

Description: 36% reduction in errors. Through the introduction of a common configuration management system within their development department, Sysdeco GIS A/S vastly decreased the number of errors in their software products prior to release, as well as reducing the time-to-market. The overall effect of this exercise was to raise the awareness of the true cost of correcting errors, as well as to reduce product development costs.

Lessons Learned: There were many lessons learned during this experiment. The main lessons were

  • To implement a configuration management system takes a long time. For the implementation to succeed, it should not be rushed. SGIS's initial estimate for the implementation was that it would take two months. In fact, it took twelve months.

  • Do not underestimate the cost of implementing and maintaining a configuration management system.

  • Do not underestimate the number of licences required (i.e., for the configuration management software). Plan on one license per developer.

  • Ensure that the hardware on which the system is to operate is powerful enough. Err on the side of excess capacity rather than not have enough, as this will influence the performance of the system.

  • Document every stage of the process followed during implementation and ensure that clear procedural documents are produced. This will establish a practical way of operating the system within the company.

Quotation from the report: "Implementing configuration management (CM) software is only part of the whole story. The major components of the whole exercise are to implement supporting CM procedures, to get CM accepted as an integral part of a company's culture, and to establish the correct measures."



Configuration Management Principles and Practice
Configuration Management Principles and Practice
ISBN: 0321117662
EAN: 2147483647
Year: 2002
Pages: 181

flylib.com © 2008-2017.
If you may any questions please contact us: flylib@qtcs.net