EuroSPI 2000 
Practical and innovation based software process improvement to prepare for the new millenium.
European Software Process Improvement
SPI and Procurement
Category Index
Rated Newspaper Supported by EU Project 

 
 
 
 
 

Software Acquisition:
Experiences with Models and Methods
 
 
 

Gerhard Getto
Thomas Gantner
Ton Vullinghs
DaimlerChrysler, Research and Technology
Ulm, Germany


 

Abstract

At DaimlerChrysler, the greatest part of the business and automotive software is provided by suppliers, often developed in complex software acquisition projects. Therefore, the development of a mature software acquisition process is essential for business success. One task of the DaimlerChrysler software process research department is the evaluation, customisation, and implementation of methods and techniques for software acquisition. This paper gives an overview of software acquisition process improvement activities that were performed at DaimlerChrysler. We used two SPRITE-S2 software acquisition initiatives (ASSIST and PULSE) as a source of information for our own generic software acquisition reference process (GARP). Furthermore, we describe how we used these initiatives as an implementation basis for our software acquisition assessment method (SAM).
 
 

Introduction

As in any other branch of industry, software is becoming an increasingly important factor for DaimlerChrysler’s business success. Over the last years several large projects, as well in automotive as in non-automotive areas, have been carried out, in which the development of technically complex software applications played a primary role. DaimlerChrysler’s core business is the development of transportation products and services. A large part of software development is being out-sourced. To fulfil its position as a premium quality producer, DaimlerChrysler has to select and manage its software suppliers in a professional way.

An important task of DaimlerChrysler’s process engineering research group is to study, develop, and evaluate relevant methods for software acquisition. Normally, the results of these research activities are tailored and transferred to the business units. As a part of these research activities we have been working on the comparison of different SPRITE-S2 software acquisition models. SPRITE-S2 is an initiative of the European Commission to support and guide in the procurement of ICT systems and services conforming to open specifications, and to promote and disseminate best practice by the application, validation and demonstration of existing new methodologies and tools. In this paper we describe how the results of the ASSIST [1] and PULSE [11] projects were used as an input and as a frame of reference for our own generic software acquisition process and assessment method.

Goal of our activities is to establish a continuous improvement process for the software acquisition processes of our internal customers. Our assessment method and generic reference process (including best practices) are important inputs for such an improvement cycle (see Fig. GGV.1):

Fig. GGV.1: Software acquisition improvement cycle

To achieve our goal, we proceeded in the following way: First, a generic software acquisition reference process including a set of best practices was developed. This process is based on an internal DaimlerChrysler acquisition process enriched with elements of the Euromethod [2]. The generic reference process was compared to the acquisition processes as described by ASSIST, and PULSE. In parallel, we performed an ASSIST-based self-assessment at our marketing-, sales- and after-sales department, thus gaining first experience on software acquisition assessments. Furthermore, additional self-assessments in different departments of the passenger car division were performed, based on the PULSE method. These experiences led to the decision to take PULSE as a starting point for our own software acquisition assessment method. We have customised the PULSE assessment method to the special needs of our business units. Currently, we are working on the integration of the assessment method and the generic software acquisition reference process.

Overview. In Section 2 we describe our generic acquisition reference process (GARP) and relate this model to the SPRITE-S2 approaches PULSE and ASSIST. In particular, we describe our criticism on these models and motivate the tailoring steps that we had to make. In Section 3 we describe the development of our software acquisition assessment method (SAM). Again, we relate the assessment method to the procedures as described in ASSIST and PULSE. Moreover, we briefly document our experiences with the application of the assessment model in our business units. Finally, in Section 4 we describe the integration of GARP and SAM.
 
 

Software Acquisition According to GARP

Motivation

Software acquisition and supplier management are a critical issue for many enterprises world-wide. Software acquisition is concerned with the provision of an organisation’s need for software. Supplier management applies to management tasks while dealing with suppliers or subcontractors. Their impact on business processes is increasing dramatically. US enterprises spend more than $250 billion per year for procuring software products and services. Software investments may result in essential productivity and customer benefits as well as in strategic advantages. Following a systematic approach for software acquisition and supplier management might help to increase the benefits on these investments.

DaimlerChrysler is purchasing more and more software products. Software acquisition and supplier management are major issues. Professional management of a customer-supplier-relationship sets a number of prerequisites, such as requirement management, risk management, well-established communication, legal aspects as well as controlling ambitions and costs. To avoid failure of a software acquisition, a strategy for procuring software needs to be in place. The elements of such a software acquisition strategy should consist of defined processes based on best practices for each phase of the acquisition lifecycle.

There are numerous existing standards and emerging approaches for software acquisition world-wide. Most often they are conflicting and a general trend for uniformity is not perceptible. A DaimlerChrysler study on models and approaches for software acquisition and supplier management [5] gives an overview of the state of the art of key processes in this area. In this survey, models, methods and best practices were described, evaluated and compared. We used the results of this survey to build a generic software acquisition reference process (GARP) and a list of references to best practices for singular sub-processes. The foundation of this software acquisition reference process was the DaimlerChrysler IT project management handbook. This handbook describes project and quality management processes and techniques. We compared our company process to the following approaches:

We detected a number of relevant sub-processes that were not covered by our company internal process. Accordingly, we extended the company internal process. The result is a comprehensive software acquisition reference process that may be used for the definition of a tailor-made business unit software acquisition process.
 

SPRITE-S2 Approaches for GARP

In this section give a brief description of the two SPRITE-S2 approaches we used as an input for our work: ASSIST and PULSE.

ASSIST deals with application of procurement process improvement techniques for small enterprises (SME). Some benefits provided are [1]:

PULSE provides a methodology for assessment and benchmarking of procurement processes. It is an organisational method for improving ICT procurement. Some main objectives of the methodology work of the PULSE project are [11]: ASSIST and PULSE follow two different strategies. ASSIST follows a bottom-up approach, that is, it helps to evaluate and improve software acquisition by defining a list of best practices and improvement recommendations. Its strength is to give hints for improvement. The goal of ASSIST is not to define and to structure a complete software acquisition process. Therefore the influence of ASSIST concerning GARP is located mainly on the level of best practices and not on the process level. We do not use the ideas of ASSIST for the evaluation and definition of software acquisition processes. Rather we use it as a source for potential improvements in our processes.

PULSE is a top-down oriented software acquisition assessment approach, that is, it uses the ideas and structure of ISO/IEC 15504 (SPICE) to define its assessment model. PULSE is the intended ISO standard for software acquisition processes. PULSE is a comprehensive description of relevant sub-processes for successful software acquisition projects and therefore an important input for GARP. We could not directly use PULSE for structuring GARP, because PULSE is a framework and not a lifecycle model. That means, PULSE does not define a logical ordering for all the contained sub-processes. The strength of PULSE is its comprehensiveness. We used PULSE as a kind of checklist for the sub-processes of GARP. Another way in which we actually use PULSE is for the assessment of existing acquisition processes (see Section 3). An important reason why we prefer PULSE above other approaches is its strong analogy with the assessment structure of the ISO 15504 standard (SPICE). Because one of our main internal customers is using SPICE oriented assessments for performing software supplier assessment, it is easier for us to communicate the main goals and underlying principles of PULSE to our customers. An issue that is still discussed is the way we can use the best practices as defined in PULSE. Regarding best practices PULSE breaks the structure of ISO 15504. PULSE defines best practices related to software acquisition also for level two and higher. This does not make sense according the philosophy of ISO 15504, because ISO 15504 only defines generic management practices for these levels.
 

The Generic Software Acquisition Reference Process (GARP)

The actual version of the developed generic software acquisition process is a hierarchical lifecycle model. From the perspective of our customers it was important to show the logical sequence of the sub-processes (if possible), as this ordering corresponds to already existing descriptions of other processes in place on the customer side and it supports the understandability of GARP. In this section we present an excerpt of this model in top-down order.

The first level of GARP contains the main processes like

These processes are further refined. For example the process "Driving the Procurement Process" consists of the sequential sub-processes shown in Fig. GGV.2.

Fig. GGV.2: Driving the procurement process

On the next level the sub-process "Procurement Monitoring and Controlling" is divided into parallel and continuous sub-processes (see Fig. GGV.3). We did not describe any logical sequence here, as the represented sub-processes possess a strong parallel and continuous character.

Fig. GGV.3: Procurement monitoring and controlling

On the lowest levels of GARP there are links from the sub-processes to best practices. The best practices provide methodological support for performing these sub-processes. The links represent the connection between what to do for software acquisition and how to perform the sub-processes. In one of our earlier research papers [7] we documented best practices for the risk management process. Fig. GGV.4 summarizes the overall structure of the activities that lead to the definition of GARP.

Fig. GGV.4: The construction of GARP

Best practices support the introduction of improvements. They are a source of possible improvements for existing acquisition sub-processes. Of course, finding and defining best practices is an ongoing activity. The strategy for building a set of best practices that support our customers is described at the end of this paper.
 
 

Software Acquisition Assessments

Motivation

Within a process improvement context, process assessment provides the means of characterising the current practice within an organisational unit in terms of the capability of the selected processes. Analysis of the results in the light of the organisation’s business needs identifies strengths, weakness and risks inherent in the processes. This, in turn, leads to the ability to determine whether the processes are effective in achieving their goals, and to identify significant causes of poor quality, or overruns in time or cost. These provide the drivers for prioritising improvements to processes. (SPICE Introduction [9])

DaimlerChrysler is getting more and more aware of the importance of software acquisition processes. In 1998, first experience with software acquisition processes were collected in the passenger car development business unit (PCD). SEI’s software acquisition capability maturity model (SA-CMM) [12] was used as a reference model for process improvement. The key process areas that we focussed on were requirements management, evaluation of intermediary and final products, solicitation, and risk management [4]. The reasons to select those areas were mainly driven by our own experiences in those fields and ad-hoc decisions of our customers. In particular, we did not perform assessments to identify strengths and weaknesses in a structured way. To bridge this gap and to better respond to the needs of our customers, we decided to develop our assessment method. Since GARP was still under construction by then, there was no assessment method available that uses GARP as a reference process model. We decided to develop a DaimlerChrysler specific software acquisition method based on already existing approaches ASSIST and PULSE.

The following steps led to the definition of our software acquisition assessment method (SAM):

  1. collect assessment experiences using ASSIST and PULSE
  2. select a basic software acquisition method
  3. customise this method to DaimlerChrysler specific constraints
  4. collect experiences by applying the customised method in pilot projects
  5. improve the customised method based on these experiences
  6. integrate SAM and GARP

Experiences using ASSIST and PULSE

The first customer for software acquisition assessment and improvement was the IT Marketing, Sales, and Aftersales Department, which accelerates the development, deployment, and support of IT solutions.

When planning the assessment process, the presence of experienced project managers is crucial. PULSE is much more comprehensive and therefore more time-consuming in its execution than ASSIST is. Because we wanted experienced project managers to participate in the assessment, we decided to use the ASSIST assessment. The assessment process and results are described in one of our experience reports [6].

First experiences with PULSE were gained in a large project in the passenger car development business unit (PCD). A special software quality taskforce was interviewed in a self-assessment workshop to analyse the current project situation, to solve existing problems, and to find pointers to areas for process improvement. The assessment results helped us to extend an already existing activity list with new, urgent activities.

The experiences affirmed our impressions that ASSIST is a pragmatic, best practice based bottom-up approach, whereas PULSE is a comprehensive, theory and best practice based top-down approach. When shortage of time of manager, project leader and project members is the biggest assessment constraint, ASSIST or similar approaches could be used to perform a "light assessment". If the assessment starts a systematic process improvement activity, PULSE is, in our opinion, a better alternative.
 

Customizing a selected method to DaimlerChrysler specific constraints

Because of its systematic, comprehensive, and SPICE oriented model, we selected PULSE as a basis for SAM. Based on experiences with ASSIST and PULSE we got process descriptions and supporting practices for two different assessment types: Some of our customer wanted to use the software acquisition assessment as a starting point for an extensive improvement program. In this case, a "light assessment" would not return enough useful information. On the other side, a "full assessment" requires a lot of resources and is only applicable in an organisation if the improvement initiative is initiated by upper management. This was not the case in our company. We decided to develop a "restricted assessment process". It is a restriction of the PULSE assessment and is based on interviews and workshops. The main differences between SAM and the PULSE assessment method are: Our actual restricted assessment process contains ten steps:
  1. Presentation of assessment approach to managers: A workshop will be held with participation of all managers of departments that will be assessed. Goal is to introduce the assessment method, get a commitment to do it, select projects and people for the actual assessment, and define a first rough assessment schedule.
  2. Identification of existing processes: A structured interview with a representative of each organisation or project who participates in the assessment is performed. One goal is to describe existing processes and/or practices and roughly classify them according to related PULSE process categories. Additionally, the interviewed people give their opinion on which processes should be assessed.
  3. Presentation of assessment approach to the assessment team: A workshop is held with all people who participate in the assessment. The goals are to introduce the assessment method, present the result of the interviews (step 2), select processes that have to be assessed, identify constraints for the assessment, and define a detailed assessment schedule.
  4. Inventory of selected projects: During the interviews in step 2 people are asked to give additional information about their project. The documents are analysed by the assessors in order to get a list of used processes, standards, techniques and best practices.
  5. Assessment interviews: Department managers, project leaders, project participants and method specialist are interviewed to disclose the strengths and weaknesses of the selected projects.
  6. Analysis of interview results: The assessors analyse the results of the interviews (step 5). The goals are to compile the strengths and weaknesses results of all interviews, to derive ratings and capability profiles, and to discover contradictions.
  7. Assessment workshop: A third workshop is held with all people who participate in the assessment. The results of step 6 are presented and discussed, especially the contradictions. The goals are to get an agreement on process strengths and weaknesses, to present ratings and capability profiles, and to select and agree in a prioritised list of processes that should be improved.
  8. Improvement planning: The assessors develop a proposal for an improvement plan regarding the most important processes. It contains a list of activities, resources, and a schedule.
  9. Presentation of improvement plan to assessment team: A final workshop is organised with all people who participate in the assessment. The improvement plan is presented and discussed. Goal is to get an agreement on this plan.
  10. Presentation of improvement plan to management team: The assessment results including the improvement plan is presented to the top management of the assessed organisation. They have to agree in the proposed improvement activities and release the necessary resources. A second goal is to get a commitment for a continuous software acquisition assessment and improvement program.
Currently, we are performing a "restricted assessment" at the marketing, sales and aftersales department. Goal of the assessment is not only to find pointers to areas of improvement for this department, but also to improve our own assessment method. The overall amount of time 16 participants have to spend is approx. 30 days. Preliminary results justify our structured medium-weight approach.

Besides gentle side-effects like creating a better awareness for process improvement issues, the restricted assessment leads to a more transparent and understandable selection of improvement areas. For example, to support the selection of the processes that have to be assessed (step 3), we used a simple portfolio technique to rank the different processes (see Fig GGV.5). We created a similar portfolio after every individual assessment interview (step 5). By comparing these portfolios we could verify whether individual answers conflict with the results of the second workshop (step 3). Conflicts will be discussed and resolved in the third workshop (step 7). Furthermore, the individual portfolios clearly illustrate which participants are experts regarding some specific process. In this way, the exchange of best practices among all participants is promoted.

Fig. GGV.5: Process prioritisation using a portfolio technique
(the numbered bullets refer to different software acquisition processes).


Integrating GARP and SAM

To have an effective overall software acquisition methodology, the scope of processes that can be evaluated by our assessment method should correspond to the sub-processes of GARP and vice versa. As mentioned before, GARP is based on an extended DC-internal life-cycle model for software acquisition. On the other hand, our assessment method is strongly related to the acquisition assessment model of PULSE. The question arises how to integrate these two models. In Fig. GGV.6 we have depicted the mapping that we have to perform for this integration.

Fig. GGV.6: Integrating GARP and SAM

In accordance with the PULSE methodology, our overall assessment methodology contains three main components:

Since GARP partially originates from PULSE, it is possible to list for every (for our company relevant) PULSE process a corresponding GARP process. On the other hand, GARP contains processes that are not covered by PULSE. For these processes we have to adapt the assessment model. For example, GARP contains a "transition to support" process, describing how to prepare and perform the transition of a developed system in its operating environment. For this process we have to define capability levels, process attributes, and best practices.

The actual state of the integration is that we have identified those processes in GARP that are not yet covered by the PULSE processes. We are currently working on the theoretical extension of the acquisition model. In parallel, we try to extend our set of best practices based on real-life experiences in software acquisition process improvement.
 
 

Conclusions

We presented our research activities in the field of software acquisition process improvement. We described a generic acquisition reference process that guides the improvement of existing acquisition processes. Pointers to areas for improvement are delivered by process assessments. We developed a software acquisition assessment approach based on the PULSE methodology. Finally, we sketched how we will integrate our generic process and assessment method.

As the integration activities are not finished yet, this will be a main topic for further research. Another future activity will be the application of the developed concepts in co-operation with our company-internal customers and the recording of these experiences. First assessments show promising results: process evaluations are performed in a well-structured and efficient manner. Especially the latter is important for our customers. Another necessary task is to extend our generic acquisition reference process with more best practices, supporting methods and facilities. The long term goal is to provide complete methodological support for our generic reference process. This means that all relevant sub-processes are covered from the process side as well as from the methodical support side.
 
 

References

[1] ASSIST: Leitfaden für die Beschaffung von IT für mittelständige Unternehmen, QAI Europe, 1999.
See also: http://www.ispo.cec.be/sprites2/fi-assist.html

[2] Euromethod: http://www.ispo.cec.be/sprites2/spriemet.htm

[3] Federal Aviation Administration: Integrated Capability Maturity Model, http://www.faa.gov/aio/icmm/FAA.htm

[4] Gantner T., Weber T.: A Way to a Comprehensive Process Model for Software Acquisition from the Viewpoint of the Customer, Proceedings of the 6th European Conference on Software Quality, 1999

[5] Getto, G.: Evaluation von Prozessen und Methoden für Softwarebeschaffung und Management von Auftragnehmern, Proceedings of the 4th Kongress Software Qualitätsmanagement "Made in Germany", 1999

[6] Getto, G., Gantner T.: Software Acquisition Processes at DaimlerChrysler AG: Research Activities for Improvement, Proceedings of the European Conference on Software Process Improvement (SPI99), 1999

[7] Getto, G., Landes, D.: Risk Management in complex Project Organizations: A Godfather-driven Approach, Proceedings of the 30th Projektmanagement Institute Conference (PMI99), 1999

[8] IEEE 1062: IEEE Recommended Practice for Software Acquisition, IEEE Standards Association, 1993

[9] ISO/IEC 15504: Information Technology – Software Process Assessment, ISO/IEC JTC1/SC7/WG10, 1998

[10] ISPL: ISPL (Information Service Procurement Library) Managing Acquisition Processes, ten Hagen & Stam, 1999

[11] PULSE: A Methodology for Assesssment and Benchmarking of Procurement Processes, http://www.ispo.cec.be/sprites2/fi-pulse.htm

[12] SEI: Software Acquisition Capability Maturity Model (SA-CMM) version 1.01, Technical Report, CMU/SEI-96-TR-020, 1996
 
 

About the Authors

Thomas Gantner studied mathematics at the University of Ulm in Germany. After obtaining a master’s degree in 1991 he joined the software process engineering group at the DaimlerChrysler Research Laboratory. Here he is employed as a research manager. He is performing research and consulting activities in the areas of software quality management and software process design and improvement, especially for software acquisition processes. At the moment he is the project leader of a long term research project with the passenger car development business unit as the key customer. Other company-internal customers he worked with are in the field of technical and administrative software. Gantner can be reached at DaimlerChrysler AG, P.O. Box 23 60, D-89013 Ulm, Germany, or by e-mail at thomas.gantner@DaimlerChrysler.com.

Gerhard Getto is a manager at DaimlerChrysler Research and Technology in Ulm, Germany. His research activities concern IT management processes and methods like risk management, quality management, software acquisition management, and supplier management. Getto has a degree in mathematics from the University of Stuttgart, Germany. He has 19 years of experience in the IT area. During this time he was active in software development, system implementation world-wide, business process analysis, information management, project management, quality management, risk management, and research. He has published papers in the areas of risk management, quality management, and software acquisition. Getto can be reached at DaimlerChrysler AG, P.O. Box 23 60, D-89013 Ulm, Germany, or by e-mail at gerhard.getto@DaimlerChrysler.com.

Ton Vullinghs studied computer science at the Catholic University of Nijmegen in the Netherlands. After obtaining a master’s degree in computer science in 1993 he joined the software methodology and compiler construction group at the computer science department at the University of Ulm (Germany). Here he was employed as a teaching and research assistant and obtained a PhD for his research in the field of functional programming and graphical user interfaces. Currently, Ton Vullinghs is working as a research assistant at the DaimlerChrysler Research Laboratory in Ulm. He is performing research and consulting activities in the areas of software quality management and software process design. His main topics of interest are software risk management and software acquisition management. He can be reached at DaimlerChrysler AG, P.O. Box 23 60, D-89013 Ulm, Germany, or by e-mail at ton.vullinghs@DaimlerChrysler.com.
 
 

Company Description

DaimlerChrysler AG is a stock company organised under the laws of the Federal Republic of Germany. DaimlerChrysler provides a wide range of transportation products and financial and other services. DaimlerChrysler is world renowned for its high quality products which reflect a long tradition of exceptional engineering, performance, service and safety. DaimlerChrysler’s Research and Technology Division is the hub of activity when it comes to the securing of the Group´s technological future. Its responsibilities are to support the business units in the development of their technology strategies, to ensure integrated innovation and technology management and to create the technological basis for differentiating products in the marketplace. The software process engineering research group is a competence centre for software processes and possesses, among other competencies, theoretical and practical know-how in the field of software acquisition management. One of the main objectives of the research group is to tailor and transfer relevant results to the business units of DaimlerChrysler.
 


 
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