EuroSPI99 
Learn from the Past - Experience the Future
European Software Process Improvement
SPI and Configuration Management
Category Index
Rated Newspaper Supported by EU Project 

Configuration Management Deployment & Practice Experiment

Silvia Mazzini
Intecs Sistemi, Pisa
 
 

Introduction

The CMEXP (Configuration Management Deployment & Practice Experiment) project is a process improvement experiment (PIE) funded by the European Community (ESSI Project 27637), conducted by Intecs Sistemi.

The overall objective of the experiment is to improve the practice of Configuration Management, by the adoption of state of art Configuration Management practices, based on a sound and ready available technology (ClearCase).

From the point of view of achieving specific measurable objectives, the target is to enhance the Intecs SPICE CM practices profile, from the current Level 1 (performed practices, informal process) "Partially Performed" mark to a "Fully Performed" Level 2 (managed process) mark and at least a Level 3 (defined and standardized process) "Partially Performed" mark.

We expect that the establishment of mature CM practices bring, per se, important benefits to the effectiveness and predictability of the overall Intecs Sistemi Software Life Cycle Process and to the quality and maintainability of its products. Furthermore, the experience gained from the PIE is expected to enhance the company know how and image, in terms of competence and user satisfaction, so far contributing to strengthen the Intecs Sistemi offer of Software Engineering consultancy and services.
 

Starting scenario

Intecs Sistemi have defined advanced software process and product models for Configuration Management (CM) in the context of a number of R&D and Industrial studies. However, these were heavily relying on highly evolved Software Engineering repository facilities provided by the Object Management System (OMS) of PCTE , an established international standard [1] that has so far failed to translate into adequately supported commercial implementations.

On the other hand, though Intecs Sistemi have achieved an overall Software Process maturity rating BOOTSTRAP Level 3.1 (defined and enforced practices) confirmed by a SPICE Assessment [5], actual deployment of Configuration Management (CM) practices scores far below that mark for projects not strictly regulated by space or defense standards.

A thorough investigation for alternatives to PCTE has identified ClearCase as a sound enabling technology for advanced Configuration Management Practices. Besides being a world class selling tool, ClearCase features capture most of the basic concepts of PCTE (attributes, links, schemes) without loosing full compatibility with Unix and binary Unix tools.

The experiment is an important step from plain Unix directories to more advanced SE Repositories oriented solutions. Integration and consistency of CM practices within Intecs Software process and product model are a primary concern.

Intecs Sistemi staff with the involvement, as consultants, of experienced SPICE (ISO IEC 15504) assessors (QUALITAL) mostly does the experiment. The Intecs Sistemi persons involved in the PIE are:

All persons above are involved part-time over the project time span; the Software Engineers are qualified computer specialists, members of the baseline project.

Company Context

Intecs main area of business is the development of system applications and the provision of consultancy services to other organizations operating in advanced technical domains such as aerospace and telecommunications. Another Intecs business sector is the development and marketing of CASE software products. The PIE has particular relevance on this business field, even if a significant degree of transferability of the experience to other business areas is expected.

Configuration and Version management are the keystone of Application Management (AM) for software artifacts that have to exist in the different temporal versions and for different environment configurations; AM has to keep track of many types of product components (code files, documentation, test procedures, SPRs, etc.) and of complex relationships (dependency, composition traceability etc.) among them. A further goal is to provide support for distributed development: Intecs Sistemi has premises in Pisa, Napoli, Roma, Piombino and Toulouse (F) and software projects may involve co-operation among teams from different sites.

Intecs have obtained ISO 9001 certification, and their software process maturity has been assessed according to the SPICE (ISO IEC 15504) and BOOTSTRAP models. These process assessment exercises have pointed out Intecs current CM practices as the weakest aspect of the Application Management process. From a qualitative point of view, an apparently disjoined overall picture has emerged:

From the quantitative point of view, and because of the above reasons, no reliable assessment of CM practices has been finalized. A "Level 1" score for CM practices has been estimated informally, without a proper CM "profile" identification (SPICE Sup-2 process).

The CMEXP PIE aims to combine Intecs experience and state-of-art technology and techniques to overcome most of the problems and deficiencies of current CM practices. In particular, we expect to achieve a fully operational demonstrator of advanced CM facility suitable to be large scale deployed as standard support for Intecs software development activities, as well as fully defined and validated CM process, procedures and guidelines, ready to be integrated into the company ISO 9000 conforming Quality System.
 

Baseline project context

The baseline project is the UmlNICE internal product development project.

UmlNICE is an integrated toolset providing support for the Unified Modeling Language (UML) [6]. It is based on state of art technology (CORBA, Java, UML) and is planned to be available in various product configurations on a number of platforms - potentially on all platforms providing support for Java and CORBA. UmlNICE is composed of a large number of components, tightly integrated among them. It can be regarded as an open framework for CASE. In addition to traditional requirements for corrective maintenance, UmlNICE is designed to support continuo product evolution for extensions, tailoring, and tracking of market demand, evolution of the method and enabling technology.

The UmlNICE project follows an "iterative and incremental" development process inspired at the Unified Software Development Process, the process that has been defined to become the standard process to use UML by the authors of [5]. The project is presently in its 9th intermediate, though self-contained, iteration, which is expected to complete by June 1999 and deliver the first commercial release.

The UmlNICE project was started at the end of 1996 and is expected to complete in the first half of year 2000; the project currently involves a team of ten persons and is planned to remain constant up to the end.
 

Experiment description

The CMEXP PIE aims at improving the contents, extent and maturity of Intecs Configuration Management practices.

The technology introduced by the experiment are the ClearCase configuration management system [8] and the ClearDDTS change request management system [9].

ClearCase is a world class selling tool, giving support for version control, workspace management, build management and process control. Integrated with ClearDDTS for the change control management, it provides a wide configuration management solution.

The ClearCase approach to the multi-version file extends the Unix and the Windows NT file system to make it a real multi-version project repository, in a transparent way. It makes it easy for an organization to deploy ClearCase, without forcing changes in the existing environment, tools, or the way of work.

Based on an intuitive web-based interface, ClearDDTS allows to track and manage both defect records and enhancement requests, can be integrated with the configuration management system and is flexible enough to be adapted to different organization needs. Particular attention has been paid to the modeling of CM processes, products, and roles and to the definition of procedure and guidelines to ensure that their implementation can take advantage by the advanced features of ClearCase.

Beside the configuration management support process, the main processes effected have been the software development, integration and testing and the whole system integration testing and maintenance, with a slight change of the configuration management, project management and developer roles.

Phases of the experiment

Beside the more obviously identified work items related to Project management, co-operation and dissemination, the core CMEXP activities are organized in two phases, the experiment preparation and execution.

The project has started in June 1998, for the duration of 18 months. At present it is in the execution phase.

In the experiment preparation the ClearCase Configuration Management System and the ClearDDTS Defect Tracking System have been acquired and installed at Intecs and an advanced course has taken place at Intecs premises in Pisa. The purpose of the course was to train the CMEXP project team, including people from the Baseline Project UmlNICE, for the use and administration of selected technologies.

In the meanwhile the current Intecs CM Practices have been assessed with the involvement SPICE assessor consultants, by setting the Initial CM Profile as a baseline reference for the measurement of the CM process improvement.

The key findings of the initial SPICE assessment have confirmed that the CM process was not completely performed at Intecs, therefore an improvement action was needed first in the activities concerning base practices, such as:

  1. the development of a general configuration management strategy, implementation and verification criteria,
  2. the establishing of a configuration management system and
  3. the recording and reporting of the status of configuration items.
As far as the capability dimension is concerned, the low rating at Level 1 was a direct consequence of the incomplete implementation of the process, while the general situation at Level 2 was slightly better than expected, due to the existence of documents modeling the various activities. Everything found at Level 3 was a direct consequence of the implementation of the ISO-9001 certified quality system at the company level.

Taking as input the results of the assessment and the current Intecs CM practices, a generic model of the CM procedures and product entities has been defined, to drive and support the correct execution of CM activities.

The activities for the experiment execution have started with the analysis of the Baseline Project UmlNICE and the tailoring of the Configuration Management System and the Generic Definitions to the baseline project specificity, by instantiating specific CM process products, procedures and guidelines.

Being composed of a large number of components tightly integrated among them, the UmlNICE project follows an iterative and incremental development process, where sometime different development activities run in parallel and the software components evolve through different versions, integrated at different project milestones.

ClearCase allows maintaining a unique repository where to collect all the software versions and releases and developing parallel versions that can be easily integrated by merge facilities; on the other hand the CM procedures become more complicated and are more demanding for the configuration manager.

As a preliminary for the exercising of CM practices within the baseline project, the specific CM guidelines and procedures have been tested by carrying out typical development and maintenance activities, in parallel to the main baseline project activities.

After the period of parallel running, the database has been reloaded with the latest UmlNICE code and the baseline project team has started the CM practices. The PIE team is currently providing guidance and support to the baseline project team and monitoring the project activities.

At the completion of the experiment feedback and lesson learned will be collected and processed and a final SPICE assessment will quantify the process improvements achieved.

Two internal dissemination events have been organized at Intecs, with the participation of the PIE and the Baseline UmlNICE projects and other internal software practitioners and managers:

Consultancy during the experiment

Consultants from the QUALITAL Consortium conduct CMEXP initial and final SPICE assessments.

The QUALITAL Consortium is a non profit institution based in Pisa particularly active in the domain of quality and Quality Systems Certification that participate in the SPICE initiative.

The training of the CMEXP project team for the use of the selected CMS and the training of one CMS Administrator for the installation of the software configuration and for related user support activities has been taught "in house" at Intecs by an ARTIS consultant; ARTIS were the ClearCase distributors for Italy.
 

Resulting scenario

Technical impact

Since the initial period, the adoption of the ClearCase configuration management system and of the tailored configuration management strategy for the baseline project has resulted in the following short-term improvements: A side effect has been also the increment of the internal problem reports, both enforced by the process and encouraged by the ClearDDTS web based interface.

Quantitative measurement of the achievement of the PIE objectives encompasses comparing the results of the initial SPICE assessment of CM practices with a final assessment planned to take place at the end of the project. The SPICE assessment results in detailed profiles of practices and allows capturing improvement in contents, extents and maturity level of CM practice.
 

Business impact

The business results of the application of a more rigorous CM process and the use of an advanced CM system in the development process are already visible as a significant improvement of the quality and efficiency of the development and maintenance processes.

The extent of the improvements, including the quality of the product and the predictability of the maintenance process, can not be quantified yet. They will become more evident when the developed product will be commercially delivered.
 

Organization impact

The adoption of the ClearCase configuration management system and the defined CM practices have forced a change in the role of the configuration manager, requiring him a deeper involvement in the project. The configuration manager becomes a sort of assistant to the project leader, in a wider range of activities, not only for the configuration management support process, such as control of the releases and of the changes, but also for the set-up and maintenance of the ClearCase environment for the development, integration, and testing processes.

We expect to transfer the PIE experience to other company development projects in the same or other application domains, therefore resulting in an overall enhancement of the Intecs Sistemi software process.

We intend to institutionalize the CM practice as part of the company Quality System.
 

Culture impact

Up to now the old Unix configuration management system SCCS was used in the baseline project. The development and maintenance are depending on manual and script procedures. Similar CM practices have been in use for many years and are well accepted. The change in the working environment has generated some skepticism and resistance initially.
 

Skills impact

The staff involved in the project has acquired an improved skill in the field of configuration management in terms of: ē greater awareness, both at the managerial and technical level, of the benefits of effective CM practices.
 

Key Lessons learned

 

Technological point of view

ClearCase is a sophisticated CM system, suitable for large and complex development activities, such as parallel developments, customizations, availability on multiple platforms, bug fixing of previous releases.

The drawbacks are that the costs of purchase, training, installation and administration are high. In particular a project should buy a license for each developer, each license costing about 4000 ECUís.

Another weak point is that ClearCase is not yet available for some of the new development platforms in use within the company, such as PCs running with Windows or Solaris x86.

To cope with such problems, Intecs Sistemi have investigated other available commercial tools and have short-listed Visual Source Safe and PVCS, as suitable to be used as complementary configuration management systems. Visual Source Safe version 6.0 emerged as the preferred option for small projects based on Microsoft Technologies.
 

Business point of view

Configuration Management is essential for a process management and control, and an optimized development and maintenance process.

The managers tend to underestimate the CM. Dissemination actions are helpful to make all the company managers and project responsible aware of the add-on and improvements that the introduction of an advanced CM technology and a more formalized process can provide.

Moreover the participation of managers is necessary to reduce the initial negative feeling, easily coming out starting with a new sophisticated CM tool and motivates the team to change the work environment and accept a more formal process.
 

Strengths and weaknesses of the experiment

CM is typically a complex issue but projects assign to it just a small percentage of the overall budget. The PIE gives the chance to analyze the whole problem and to find proper solutions, applicable also to future projects.

A weakness of the experiment is that the selected CM technology is quite sophisticated, but also expensive. It may result that many projects, having a moderate degree of complexity and a small development team, adopt cheaper solutions, such as Visual Source Safe or PVCS.
 

Conclusions

We expect to be able to demonstrate the benefits of a more rigorous and formalized CM process and of the use of an advanced CM tool that allows a better representation of product characteristics and the enforcement of the integrity constraints. Having identified the CM process fragment as the weakest point of the overall software development process, we expect a significant improvement of the software life cycle process as a whole, in particular with respect to Application Management activities.

However, CASE product development and maintenance, which is the scope of the baseline project, is among the most demanding class of activities with respect to CM; experiences and achievements can easily be transferred to other fields. In particular, a successful demonstration allows transferring the results of CMEXP to most of Intecs Sistemi software development projects in a number of other application domains.

Achievement of the objectives of the experiment might therefore result in an overall enhancement of reliability and efficiency of the Intecs Sistemi software process, of the company know how and of its image (competence and user satisfaction). All these aspects are important factors affecting the competitiveness of the company.
 

References

[1] Baudier G., Gallo F., Minot R., Thomas I., An overview of PCTE and PCTE+, in: Proceedings of the 3rd ACM Software Engineering Symposium on Practical Software Development Environments, pp. 107-109, ACM Press, USA, 1989.

[2] Boehm B. W., Software Engineering, IEEE Transactions on Computing, Vol. 2, pp. 1226-1242, 1976.

[3] Boehm B. W., A Spiral Model of Software Development and Enhancement, IEEE Computer, Vol. 21, N.5, pp. 61-72, 1988.

[4] IEEE 1042, IEEE Guide to Software Configuration Management Plans (ANSI), The Institute of Electrical and Electronics Engineers, USA, 1987.

[5] ISO/IEC 15504, SPICE Ė Software Process Assessment, ISO/IEC Copyright Office, Switzerland, 1998

[6] Rumbaugh J., Jacobson I., Booch G., The Unified Modeling Language Reference Manual, Addison-Wesley, USA, 1999.

[7] Rumbaugh J., Jacobson I., Booch G., The Unified Software Development Process, Addison-Wesley, USA, 1999.

[8] Rational Software Corporation, ClearCase Reference Manual, Version 3.2, USA, 1998 .

[9] Rational Software Corporation, ClearDDTS Userís Guide, Version 4.1, USA, 1998
 
 

Intecs Sistemi


INTECS Sistemi is a Software-House providing leading-edge technological support to major European organisations in the design and implementation of complex electronic systems. It operates at the forefront of the software market, where innovation, complexity and quality aspects are essential to determine the company success.

During almost 25 years of activities, INTECS Sistemi has achieved extensive experience in the production of software systems as well as software engineering and quality. Such experience has been acquired through a well-established co-operation with most of the major Italian and European electronic industries and the development of proprietary products.

Silvia Mazzini

Silvia Mazzini received her degree in computer science at the Pisa University, Italy, in 1983. From 1983 to 1987 she was with Systems & Management, Pisa, where she was involved both in industrial software development and research projects. In 1987-1988 she was in Delphi, Lucca, , where she was involved in research projects. In 1988 she joined Intecs Sistemi, where she is now senior software consultant.

Her research interests are in the field of software engineering, specifically in advanced software engineering environments, process modeling, and object-oriented technology.


Partners in EuroSPI

Editors
ISCN LTD, ISCN GesmbH, Schieszstattgasse 4/24, 8010 Graz, and Coordination Office, Florence House, 1 Florence Villas, Bray, Ireland, office@iscn.at, office@iscn.com, office@iscn.ie, Editing Done: 19.7.2002