IT-

REPORT

TitleDateIT-

Industriens

Samarbeidsplan,Bransjeutvikling

Configuration Management:Introducing procedures in two industrialcompaniesAuthor(s)1995-05-02Pages20Report nr.Rune Christensen (TrioVing),Joe Gorman (SINTEFDELAB),Sissel Kolvik (Stentofon)P-1211-13ClientClients ref.ITF/NTNFSummaryThis report describes the work of SISU project 1211 “Introducing Configuration Management Proceduresin two industrial companies”. The participants in the project were Stentofon and TrioVing (both industrialcompanies) and SINTEF DELAB (research institute).The goal of the project was to introduce concrete, practical procedures for configuration management inthe two participating industrial companies.The reports starts by describinggeneral background: earlier work in SISU (especially the generic config-uration management methodology), and existing configuration management practices and tools at Sten-tofon and TrioVing.A chapter onProject History explains how this introductory phase was followed by a detailed analysis ofprocedures, followed by refinement of these in close co-operation with staff in the companies. The pro-cedures were expressed inConfiguration Management Plans, whose gradual introduction started duringthe final stages of the project.The report also summarises theresults of the project (including documents, seminars and attitude chang-es), and future plans within the companies.Finally, the report contains conclusions from both participating companies, summarising theexperiencesgained and offeringadvice to others who wish to introduce configuration management procedures.KeywordsConfiguration ManagementQuality AssuranceMethodsReport responsible:Geir MelbyProject LeaderAddress:

SISUc/o NRBoks 114 Blindern, N-0314 Oslo 3Norge

Phone:Fax:E-mail:22 85 25 00 - 22 85 26 1322 69 76 60

geir.melby@nr.no

Rep.No.:P-1211-13II

SISU II

Table of ContentsPageTable of Contents................................................................................................................................. II1Introduction.......................................................................................................................................... 11.11.21.31.41.523Background............................................................................................................................ 1Project Goals......................................................................................................................... 1Project participants................................................................................................................ 2Starting point: Stentofon........................................................................................................ 3Starting point: TrioVing........................................................................................................ 3Project History...................................................................................................................................... 5Results.................................................................................................................................................... 73.13.23.33.4Achievement of Goals........................................................................................................... 7Documents............................................................................................................................. 8Seminars.............................................................................................................................. 10Changes in attitudes............................................................................................................. 124The way forward................................................................................................................................. 134.1Stentofon.............................................................................................................................. 134.1.14.1.24.1.34.1.44.1.54.1.64.1.74.1.84.2What we already have accomplished.................................................................. 13Training the Organisation................................................................................... 13Writing a Configuration Management Plan for the “Test Project”.................... 14Establish a database for registration of dependencies........................................ 14Establish version control for PC files................................................................. 14Reporting status for configuration management................................................. 14Implementing Configuration Management on new projects.............................. 14ISO certification.................................................................................................. 14TrioVing.............................................................................................................................. 155Conclusions......................................................................................................................................... 165.1Role of SISU........................................................................................................................ 165.1.15.1.25.2Stentofon............................................................................................................. 16TrioVing.............................................................................................................. 17Experience with introducing Configuration Management.................................................. 175.2.15.2.2Experience at Stentofon...................................................................................... 17Experience at TrioVing....................................................................................... 185.3Advice on introducing Configuration Management............................................................ 19Bibliography........................................................................................................................................ 20Table of Contents

Rep.No.:P-1211-131

SISU II

1

1.1

Introduction

Background

In the course of the SISU I project (1988-1992) configuration management (CM) wasidentified as being an important area, contributing significantly to productivity in softwaredevelopment.

Initial work on fairly brief tool evaluations[1] was followed by development of a config-uration management “scenario” [3] to be used a basis for more thorough tool evaluations.This was followed by a comprehensive, long-term assessment of the PCMS tool at Sie-mens [2].

Later work involved integration of configuration management tools with developmenttools [4], and approaches to the gradual improvement of configuration management prac-tices in companies [6]. There was also a realisation that improvements in configurationmanagement require the active support of managers in a company, and accordingly a“management seminar” was arranged to highlight the importance of configuration man-agement [5].

Towards the end of SISU I, the focus moved away from tools towards amethodology forconfiguration management. This resulted in a comprehensive document being produced[8], describing in some detail a general methodology for configuration management. Acourse on this methodology was also arranged [7].

When the SISU II project started in 1993, it was decided that it would be worthwhile tocontinue the work on configuration management methodology. The methodology [8]developed in SISU I was regarded as being a good starting point, but as being rather toogeneral to bedirectly applicable in a specific industrial context. It was therefore decidedto start a project whose main aim would be toreally introduce a configuration manage-ment methodology in an industrial setting.

The result of this decision was the project whose main results are here described: SISUproject 1211 “Introduction of Configuration Management at Stentofon and TrioVing”.The project started in May 1992 and lasted until December 1995, and the remainder ofthis report summarises the goals, experiences and results arising from the project.

1.2Project Goals

Main goal:To introduce an effective method for configuration management into the par-ticipating companies, where a methodology is considered “effective” if it supports:•rapid yet safe changes in products•considerable variety in the product base•a large number of products and customers

Introduction

Rep.No.:P-1211-132

SISU II

In addition to this main goal, several sub-goals were identified:Sub-goal 1:

The configuration management methodology developed in SISU I [8]should be customised to the needs of the participating companies.

Sub-goal 2:The relevant requirements in ISO 9001/9003 should be taken into account

in order that the companies can be ISO certified.Sub-goal 3:The methodology will be adopted within the companies, with enthusiasm

from management, the quality assurance manager and system developers.Sub-goal 4:The experiences gained in this project will be effectively communicated to

other companies participating in SISU, in order that they can benefit fromthe lessons learned.

1.3Project participants

Two industrial companies (Stentofon & TrioVing) and one research institute (SINTEFDELAB) were involved in the project. From each of these companies there was just oneemployee directly involved in the project. However, in the later stages of the project it wasappropriate to widen participation in the project within the industrial companies, as detailsof new procedures were discussed and introduced.

Both industrial companies develop products which consist of both hardware and softwareelements: Stentofon develop customized internal communications systems, whileTrioVing develop access control systems for large buildings, ships etc. Both companiesalso have many variants of their products, and so configuration management is especiallyimportant for them. It was considered advantageous to havetwo industrial companies inthe project because:

•This opens up possibilities for mutual assistance in recognising and solving commonconfiguration management problems.•Although the two companies have some things in common, there are also things thatareunique to each. Thus, the results of the project are more general, and more likely tobe applicable at a later stage in other companies participating in SISU.Therole of the industrial companies within the project can be stated quite simply: to findout what configuration management support is really needed within the companies,develop a suitable solution and implement it.

The role of SINTEF DELAB (the research institute) in the project was to:

•offer technical assistance on configuration management in general, and, in particular,customization of the SISU I configuration management methodology•assist (through participation in events such as seminars etc.) in the process of convinc-ing people within the industrial companies of the importance of configurationmanagement•assist in the production of project deliverables

Introduction

Rep.No.:P-1211-133

SISU II

Nothing more is said about SINTEF DELAB in the remainder of this report; the reportfocuses on the two industrial participants. As the situation within the two companies wasrather different, the process of introducing configuration management also proceededrather differently. As a result, most sections of this report include separate sub-sectionsfor Stentofon and for TrioVing. An alert reader may notice some overlap in some cases,and may feel that a careful document editor might have taken the trouble to state these ina common section. However, the policy which has been followed is that it is more impor-tant that the reader gets a clear picture of what happened in each of the companies; so itis better that the accounts from each company are kept independent. But it must bestressed that in the course of project both companies co-operated rather closely, compar-ing problems and devising solutions.

1.4Starting point: Stentofon

At STENTOFON the following assumptions were the basis of the project:

•There was no room for a large investment in an overall configuration management tool.The main effort should be put into establishing useful procedures with support of thetools we already had. New tools should be recommended / introduced only where abso-lutely necessary.

The tools we already used were:

—IMPCON - Database for material lists for all hardware

—SID - Identification database (locally developed add-on to IMPCON)—Database for registration of all Change Requests and Change Orders (locally

developed)—RCS Revision Control System for version control of software files—Make - tool for controlling software compilation and linking

—Softlib - Database for released and delivered software versions (locally

developed)•We had a fairly good change control system for hardware. We did not want to changethis dramatically•We were not going to implement Configuration Management on all old products. Thescheme was limited to new products delivered after 1.1.95•We did not want to limit the project to software only. The whole product was to be con-sidered, including hardware, software and documentation.

1.5Starting point: TrioVing

The main goal for TrioVing regarding configuration management has been to control oursoftware products. TrioVing develops hardware, mechanical, and software products, butthe main problem area for configuration management has been software products. The

Introduction

Rep.No.:P-1211-134

SISU II

mechanical and electronics development departments already had a locally developedsoftware tool - PUMA - handling hierarchies of parts. The tools shows, for each part,where it is used and which sub-parts it consist of.

The idea is that each software product will be registered in PUMA, but the building of thesoftware product is to be handled by the Software Configuration Management Plan(SCMP).

To ensure proper building, the following had to be defined:•Where to store documents•What documents to store

•How to document configurations and builds•How to handle changes

•Who should have what responsibilities

We had very limited defined procedures and documents to build on; almost all had to bedefined from scratch.

We decided that the best way to go was to select one specific project and write a SCMPdocument for that. We chose the VingCard Access project: its goal is to generate the nextmajor version of this access control system.

At an early stage in the project we looked at tools that might help us, but it soon becameclear that commercial restrictions within the company did not leave room for any majorinvestment in tools. So it was clear that we would have to continue using the tools wealready had, and just add manual procedures (which we might automate at some laterdate).

Summarising, our starting point consisted of some limited procedures, and the followingtools:

•Sourcerer's Apprentice - a version control system by Solution Systems. It ensuresthat the same source code is not modified by two developers at the same time. In addi-tion one can name a set of source code files as they looked at a certain point in time.•PUMA - handles hierarchies of parts (as mentioned above).

Introduction

Rep.No.:P-1211-135

SISU II

2Project History

The project started in April 1993 and formally terminated in March 1995. Even after thisformal termination, work on refining configuration management procedures is expectedto continue within the industrial companies for at least a further six months.The work of the project was divided into four main phases:Phase 1:Introductory (April 93 - August 93):

In this phase both industrial companies evaluated existing configuration managementpractices within the companies. This work involved analysing procedures and tools in use,and some assessment of the degree to which these were satisfactory.

As both companies are interested in ISO certification, ISO 9000 documents were studiedin order to form a clear picture of the requirements on configuration management imposedby the standard: this was viewed as having an important impact on the direction of furtherwork.

Special attention was paid to the issue oftool support. Economic constraints in both com-panies precluded any major investment in new tools, and it was therefore decided that thework of the project should concentrate on methodological aspects of configuration con-trol, supported in the main by tools already to be found in the companies.Phase 2:Method investigation (August 93 - December 93):

Having decided in phase 1 to focus on methods and procedures, attention in this phase wasdevoted to finding out what the elements of a configuration management methodologyshould be.

The starting point for this work was the generic configuration management methodologydeveloped in the SISU I project[8]: this was studied in detail with a view to customisingit for the needs of Stentofon and TrioVing. In addition, recommendations from the IEEE(as described in [9]) about configuration plans were studied in order to plan how best tostructure and present the plans to be developed.

On completion of this phase, project participants had a fairly clear idea of what proceduresneeded to be defined, and how these should be integrated with other existing quality man-agement procedures; they were in a position to start involving other colleagues in thework.

Phase 3:Configuration plan development phase (January 94 - July 94)

The main work of this phase was to develop initial versions of the configuration manage-ment plans for the two companies. At this stage both companies were able to work ratherclosely on agreeing the structure of the plans; in some cases it was even possible to havesome of the content in common too.

It was also in this phase that the first steps were taken to convince management of theimportance of configuration management, and to sketch out for them what the practicalconsequences would be of introducing configuration management procedures. Someresistance from management had been expected, but it turned out that management were

Project History

Rep.No.:P-1211-136

SISU II

already convinced of the benefits of configuration management, and were very supportiveof the work of the project. Nevertheless, it was clear that introducing the procedures to bedetailed in the configuration management plans would have associated costs, and manage-ment were keen to minimise these.Thus, good contact with management was established at this stage andworking groupswere established1 to examine in detail the practical aspects of introducing new configura-tion management procedures. The keyword here wasdialogue: there was no attempt byproject participants to ““preach” to colleagues about how to improve configuration man-agement; rather, some proposals were made which were then further developed in co-operation with colleagues.Phase 4:Configuration plan refinement and introduction (August 94 - Dec. 94)The final phase of the project concentrated on refining the configuration managementplans and initiating their practical introduction in the two companies. During this phase,Stentofon and TrioVing worked rather more independently; this was natural as the levelof detail of the work was more finely tuned to the specific requirements of the twocompanies.An important aspect of the work during this phase waspragmatism: it was necessary insome cases to adjust certain procedures so that they could more readily be integrated withexisting practices. Sometimes this meant accepting less than perfect solutions to technicalproblems - but ease of acceptance was regarded as being more important than technicalperfection.At Stentofon a seminar was arranged2 in November 1994,the purpose of which was topresent and discuss the (more or less) final version of the configuration management planto be introduced at the start of 19953. This proved to be a very useful exercise: participantsfrom different departments within the companies were able to point out various areaswhere the plan needed further refinement, while at the same time becoming more familiarwith the overall context within the company.At the end of this phase both companies had defined a reasonably complete configurationmanagement plan, achieved consensus about its details and come a considerable way inits practical introduction. However, in both companies the need for further refinement -based on experience after some months use - was clearly recognised. Furthermore, therewas commitment to devote effort to these activities even although this project was for-mally complete.1. Such groups were established only at Stentofon.2. There was no equivalent seminar arranged at TrioVing: it was not considered appropriate as thenumber of people affected by the configuration management plan was relatively limited.3. An introduction was also given (by Joe Gorman, SINTEF DELAB) presenting configurationmanagement in general, summarising its benefits and different technical aspects.Project History

Rep.No.:P-1211-137

SISU II

33.1ResultsAchievement of GoalsThe project goals are listed in section 1.2 on page 1. Table 3.1 summarises the extent towhich each of these goals has been fulfilled.Table 3.1 Achievement of goalsGoalsMain goalShortdescriptionIntroduction ofeffectivemethod forconfigurationmanagementExtent to which the goal was achievedWe consider this main goal to have been achieved.Both companies now have effective configurationmanagement procedures in place and - even thoughfurther refinement of these is required - productdevelopment will be considerably more effective as aresult of this work.We did not directly “customize” the SISU I CM meth-odology, but rather used it as a basis (together withother documents and knowledge of conditions withinthe companies) to produce methodologies tuned to theneeds of Stentofon and TrioVing.We are confident that the procedures introduced bythis project will be sufficient for ISO certification.(Note: TrioVing has already been certified).It would be an exaggeration to say that managementare “enthusiastic” about configuration managementprocedures - they have other concerns which occupy agreater share of their attention. But in both companiesmanagement - and persons responsible for qualityassurance - have been convinced of the importance ofthe configuration management procedures, and fullyaccept that they shall be implemented in practice.This goal has been partly achieved in that a seminarwas held with Norsonic (see section 3.3) at which theexperiences from Stentofon and TrioVing were effec-tively communicated and discussed. However, fullachievement of this goal would require:•communication with a wider range of firms•a delay of (say) six months after the formal conclu-sion of this project - in order to allow reporting ofprogress over timeSub-goal 1Customizationof SISU ImethodologySub-goal 2ISO 90001requirementsSub-goal 3Adoption withenthusiasmfrom managersSub-goal 4Experiencescommunicatedto other SISUcompaniesResults

Rep.No.:P-1211-138

SISU II

3.2DocumentsTable 3.2 : Documents producedTable 3.2 lists the documents produced in the course of the project:IdentifierP-1211-0aTitleConfiguration management:Tools and Methods in use,Spring 1993Summary of ContentsThis document describes the configu-ration management support in usewithin the participating companies atthe start of the project.It formed a baseline for further workin the project.P-1211-1How to relate to ISO 9000-3This document essentially provides asummary of the requirements appear-ing in ISO-9000 which pertain toconfiguration management.An important part of the document isa “checklist” of issues which must beaddressed by a configuration manage-ment method.P-1211-2Tool OverviewThis report describes the plans in theparticipating companies regardingadoption of tools for configurationmanagement.It is based largely on the tools alreadyin use in the companies, as economicconstraints prevented major invest-ment in new tools.This led to an important focus in theproject: onprocedures rather thantools.Results

Rep.No.:P-1211-139

SISU II

Table 3.2 : Documents producedIdentifierP-1211-3TitleConfiguration Plan, V1Summary of ContentsThis document is the result of the firstattempt to define aconfiguration planfor each of the participating compa-nies.b Work on developing morerefined versions of the configurationplans became the main focus of theproject in later stages.The plans include company andproject specific information on:•The role and importance of config-uration management within theorganisation•The existing organisational struc-ture, focusing on who has/will haveresponsibility for configurationmanagement issues•Methods for identification of itemssubject to configuration control•Change control procedures•Procedures for product release•Other procedures which are relatedto configuration management (e.g.product development, maintenance,sales and marketing etc.)•Guidelines on tool supportP-1211-5Software ConfigurationManagement DictionaryIn order to achieve full acceptance forconfiguration management in organi-sations, it is important that a fairlyprecise terminology can be used.We therefore produced this “Diction-ary” to provide a precise definition ofconfiguration management terms.It is based largely on IEEE standard610.12.Results

Rep.No.:P-1211-1310

SISU II

Table 3.2 : Documents producedIdentifierP-1211-6TitleConfiguration ManagementPlan, V2 - StentofonSummary of ContentsThis is a refined version of the Stento-fon part of the ConfigurationManagement plan. It was producedafter discussion with key peoplewithin the company, centred round thefirst version of the plan and detailedknowledge about existing practice.Final version of Stentofon’s Configu-ration Management plan, producedafter full review of earlier versionswith key people in the company, andafter the configuration managementseminar (see section 3.3).Final version of TrioVing’s Configu-ration Management plan, producedafter full review of earlier versionswith key people in the company.This report.P-1211-11Configuration ManagementPlan, V3 - StentofonP-1211-12Configuration ManagementPlan, V2 - TrioVingP-1211-13Configuration Management:Introducing procedures intwo industrial companiesa. This form of identifier is that used within the SISU project to identify deliverables.b. In this initial version, there was very close co-operation between Stentofon and TrioVing andit was easiest to produce a single document. The document did contain separate parts for eachcompany, but the overall structure was very similar. In later versions - as our ideas became moremature and the different requirements of Stentofon and TrioVing became more apparent - it wasmore natural to produce separate documents, varying more in structure.3.3SeminarsIn addition to written documents, two seminars were prepared and presented in the courseof the project - both in late Autumn 1994.The first one was a half day seminar held at Norsonic in Oslo. Norsonic had identified con-figuration management as being an area of importance to them, and - through theirparticipation in other SISU projects - had been made aware of the aims and activities ofthis project. They were therefore interested to find out more about the experiences in thisproject, with a view to considering what further actions they themselves would take in thisarea. They were particularly attracted by the fact that very concrete, company specific pro-cedures were being developed and introduced.Results

Rep.No.:P-1211-1311

SISU II

The seminar consisted of four parts:

•A general overview of configuration management problems, approaches and benefits(by Joe Gorman, SINTEF DELAB).•A description of the work being carried out at Stentofon (by Sissel Kolvik, Stentofon).This presentation focused on procedural and organisational issues - including socialaspects of introducing changes.•A description of the work being carried out at TrioVing (by Rune Christensen,TrioVing). This presentation focused rather more on technical issues, including adetailed description of proposed mechanisms (a mixture of tool support and agreedprocedures) for version control.•An interactive discussion involving all participants in this project and all members ofthe software development team at TrioVing.The two industrial presentations complemented each other rather well, and Norsonicfound the seminar useful for their future planning.

The second seminar was held at Stentofon; it lasted a full day and was different in naturefrom the Norsonic seminar. The purpose of the seminar was to ensure final agreement onthe configuration management plan whose introduction was scheduled to take place a sorttime after.

The seminar was attended by 8 senior staff at Stentofon, representing different depart-ments within the company. The daily working practices of each of the departments wouldbe affected by the procedures and guidelines described in the configuration managementplan, and it was important to be certain that the proposed procedures were acceptable andpractical. Individual discussions with each of these people (and others not present at theseminar) had already taken place in the months preceding the seminar, but it was feltimportant to use the seminar as a kind of final summary. This was particularly importantto give everyone the overallcontext - including how procedures would operate inotherdepartments andbetween departments.The seminar consisted of four parts:

•A general overview of configuration management problems, approaches and benefitsThis was given by Joe Gorman, SINTEF DELAB, and was essentially a repeat of thepresentation given at Norsonic.•A detailed presentation of the proposed configuration management plan for Stentofon,by Sissel Kolvik, Stentofon. This presentation also included a “scenario”: a concreteexample showing exactly how the proposed procedures would be applied in the caseof an imaginary (but realistic) concrete example.•A detailed discussion of the proposed plan, involving all participants.

•A final discussion to agree what changes were needed in the plan before it could beintroduced, and to consider some longer term issues.The seminar proved to be very useful: various misunderstandings were cleared up, defi-ciencies in the proposed procedures were identified, improvements were suggested andplans agreed for the detailed stages of introducing the procedures.

Results

Rep.No.:P-1211-1312

SISU II

3.4Changes in attitudes

The least tangible results of the project (but probably the most important in the long term)were the changes in attitude that were brought about amongst staff at Stentofon andTrioVing.

Before the project, there was general acceptance that configuration management was, ofcourse, important. But there was some lack of true commitment todo anything about it.This project, however, had introduction of configuration management procedures as itprincipal aim, allocated the time and effort to fully analyse the detailed problems andpotential benefit, and was able to propose practical solutions. As a result of this, staffwithin both organisations were able more readily to accept that solutions to configurationmanagement problems were beneficial, feasible and realisable. Thus, their attitudechanged from “important - we will do something about it some time” to: “important - startintroducing/improving procedures today”.

Of course, this is only a very simplified summary of a whole range of subtle changes inattitudes in a wide range of different people. But the importance of changes in attitudesshould not be underestimated: it was a crucial factor in the success of this project.

Results

Rep.No.:P-1211-1313

SISU II

4The way forward

On completion of the project, both participating industrial companies considered that theyhad achieved a lot, and made considerable progress. However, work on producing theconfiguration management plans also served to highlight areas where further detailedwork would be required. But one of the important effects of the project was to achieve aconsiderable degree of commitment from management, and there is therefore acceptancein both companies that work on refining and introducing configuration management pro-cedures should continue beyond the termination of this project.

4.1

4.1.1

Stentofon

What we already have accomplished

At Stentofon the following work has been accomplished during the project.We have:

•Established a Configuration Management Plan that has become a part of the QAsystem.•Established four new procedures - which have now been included in the QA manual -and modified some of the existing procedures•Introduced Configuration Management as a specific topic into one of our largestprojects•Changed the methods concerning numbering and versioning•Included software and documentation in the change control system•Established a method for registration of dependencies

In the time to come, we have to start using the new routines and methods. There are alsoa number of tasks concerning configuration management that have been defined and rec-ognised within the project, but which have not been fulfilled. The main tasks for 1995 arelisted in the sub-sections which follow.

4.1.2Training the Organisation

The new methods and procedures have to be introduced into the organisation in a struc-tured way. We plan to have a training session with the project members of the of theproject that has been appointed as a “test case”. One person has been assigned the respon-sibility of Configuration Management in this project.

Later we plan to have similar training sessions for our engineering staff and for the rest ofthe organisation.

This is planned to be completed in the course of 1995.

The way forward

Rep.No.:P-1211-1314

SISU II

4.1.3Writing a Configuration Management Plan for the “Test Project”

One of the new procedures states that each project shall establish its own ConfigurationManagement Plan. This plan shall be based on the overall Configuration ManagementPlan for the company. Writing such a plan for the actual project still remains. This planwill be an appendix to the procedure and can be used as a template for other projects.This is planned to be fulfilled during the first half of 1995.

4.1.4Establish a database for registration of dependencies

The basic method for registration of dependencies has been established. We need to estab-lish a simple database to provide practical support for this. This should be finished in the course of 1995.

4.1.5Establish version control for PC files

We already have good version control for UNIX files (RCS). We need to evaluate and pur-chase a good system for PC files.

This should be implemented into the organisation during 1995.

4.1.6Reporting status for configuration management

In the Configuration Management Plan, we state that there should be a way of telling theorganisation the current configuration status for a product. We have not yet spent muchtime considering how the information available in our databases should be presented tothose not deeply involved in the project. There is a need for engineering, sales and mar-keting staff to have an easily understandable presentation of the dependencies between thedifferent parts of the product.

We will look further into this in 1995, and should be able to present a solution to this prob-lem before 1996.

4.1.7Implementing Configuration Management on new projects

It will take some effort to make sure that the new procedures and methods are applied onnew development projects. One person will be appointed to focus on the matter and assistthe project participants.

4.1.8ISO certification

The company is going to be certified according to the ISO 9001 standard. The QA depart-ment aims at fulfilling this before summer 1995.

The way forward

Rep.No.:P-1211-1315

SISU II

4.2TrioVing

The first version of the SCMP for the VingCard Access project has been written and isalready being used. However, we expect that this document - and the documents to whichit refers - will change as we gain experience.

Once the SCMP has become reasonably stable, it will be generalised, and guidelines writ-ten for creating SCMPs for other projects.

The way forward

Rep.No.:P-1211-1316

SISU II

5

5.1

5.1.1

Conclusions

Role of SISU

Stentofon

Introducing Configuration Management into an organisation that has a set of establishedroutines, long traditions and a complex product spectrum is no simple and easy task. It iseasy to get lost in all the different aspects of the matter, easy to get overwhelmed by thetask and easy to say that “this is all very well, but I don’t think it will work for us”.It is unlikely that we would have come as far as we have without being a part of the SISUproject, and STENTOFON has had considerable benefit from this project.The main points to mention are:

•Configuration Management has been put on the agenda and we were allowed to spendtime on the project•The report from SISU I on “Metodikk for produktforvaltning” was helpful in the start-up period, where we struggled to define and limit the work.•In the start-up period it was necessary to have other people as discussion partners. Ithas been very useful to have people from different companies with different back-ground in the project. This added valuable input during the work and made it easier toget a good understanding of the issue and the problems and how to get on with thework.•We never lost sight of the more general aspects of the matter. That could easily havehappened if we had worked by ourselves.•Taking part in the SISU project made it easier to establish a platform inside the com-pany for “Configuration Management thinking”. Once the commitment had been madeto join the SISU project, management was very eager to see that there were someresults coming out of it.•When it came to the details in the work with the Configuration Management Plan, theSISU project could not contribute with “ready-made” solutions suitable for our organ-isation. You had to discuss and decide matters within your own organisation. AttendingSISU meetings made you sharpen your attention to the matter, and it contributed pos-itively to concentration and motivation.

Conclusions

Rep.No.:P-1211-1317

SISU II

5.1.2TrioVing

SISU has played an important role regarding configuration management for TrioVing. Itis difficult to prioritise work concerned with defining methods when our primary goal isto generate software products. SISU has here functioned as a driving power:

•Acceptance for the project at TrioVing was amplified because SISU partly financed theproject.•Milestones have been defined by SISU “independently” of what we were working atTrioVing.Other advantages have been:

•Resources outside TrioVing have been available through SISU, most importantly theother project participants.•We have gained insight into what is going on, and how things are done, in other firmsdoing software development.

5.2

5.2.1

Experience with introducing Configuration Management

Experience at Stentofon

Our experience at Stentofon can be summarised under the following headings:•It is a complex issue

Establishing configuration management involves a lot of people. Methods and routinesneed to be looked at and probably changed. Changes must be discussed and evaluatedand you have to consider the impact of the changes for all parts of the organisation.This means that you must know how things have been done - and why - to be able tosuggest good solutions.•You need to co-operate with all parts of the organisation

At Stentofon we decided to include the whole product in the configuration manage-ment activity. This meant that you had to discuss the matter with all parts of the

organisation. The need to establish responsibilities for the different aspects of config-uration management became obvious. In an everyday situation with a lot of otherthings to do, it was necessary to sell the idea of configuration management thinking tothe whole organisation. Support from the management level in all parts of the organi-sation was vital to get on with the work.•It takes time

As the changes involved different parts of the organisation and a lot of people, youneed to spend a lot of time finding out how things have been done and why. You needto know this before you can suggest any changes. You need to spend a lot of time talk-ing to people, selling the idea of configuration management and making them

Conclusions

Rep.No.:P-1211-1318

SISU II

contribute with ideas and suggestions for improvement. This is very time-consuming- but should not be considered a waste of time. The organisation needs some time todiscuss and accept the changes.

•Decisions must be made

There is the danger, however, that discussions may kill the issue. There must be a limiton who you involve and how far to stretch the discussions. Decisions have to be madeeven if the solutions are not perfect or complete.•No ready-made solutions are at hand; each company has its specific problems

You have to know what configuration management really means, and then you have tolook at your own organisation and find out what to do.

Depending on existing routines and methods, organisation structure, complexity of theproducts and available resources, the implementation of configuration managementwill be different from one company to another.

5.2.2Experience at TrioVing

A difference between Stentofon and TrioVing is that the impact of this project was muchmore restricted at TrioVing. As mentioned in the introduction, the project was onlyintended to implement configuration management for software development at TrioVing.For executables, the old system (PUMA) was considered able to handle the organizationof this in the context of the complete system. This project, then, focused on handling ofsource code and documents.

Furthermore, because of the partitioning of the software development department atTrioVing into two sections, we had to restrict most of this project to just one of thesections. This meant that the work affected mostly one manger and three developers.We expect that - as the system matures - it will be expanded to include both of the softwaredevelopment sections.

Conclusions

Rep.No.:P-1211-1319

SISU II

5.3Advice on introducing Configuration Management

Based on the experience of Stentofon and TrioVing in this project, we recommend thatcompanies wishing to introduce configuration management procedures within theirorganisation:

•remember that management support is essential

•involve all parts of the organisation -sell configuration management thinking•limit the task from the beginning

•ensure that routines survive changes in the organisation

•take care that results are so detailed that people understand what has to be done inpractice•do not ignore training the organisation - it is very important•use external resources where possible for discussions and advice•don’t give up hope!

Conclusions

Rep.No.:P-1211-1320

SISU II

Bibliography

Note: This list of references does not include documents produced as a result of the workthe project itself; these are described in detail in section 3.2 on page 8.

[1]

Terje V. Arnesen

Evaluering av ProduktforvaltningsverktøySISU rapport SISU-88008

O. Andreassen

Evalueringsresultater PCMS - SISU Configuration Managment ScenarioSISU rapport SISU-89009

Joe Gorman and Svein HallsteinsenConfiguration Management ScenarioSISU rapport SISU-89002

Svein Hallsteinsen

Integration of Configuration Management Tools with Development ToolsSISU rapport SISU-90013Arne Ramstad et al

Lederseminar i produktforvaltningSISU rapport SISU-90002

Ole Solberg

Skrittvis forbedring av konfigurasjonsstyringSISU rapport SISU-91003

Ole Solberg

Kursmateriale for “Metodikk for produktforvaltning”SINTEF DELAB / SISU 1993-08-26Ole Solberg et al

Metodikk for ProduktforvaltningSISU rapport SISU-92007

IEEE

IEEE GUide to Software Configuration MangementANSII/IEEE Std 1042-1987

[2]

[3]

[4]

[5]

[6]

[7]

[8]

[9]

Bibliography