LEGACY SYSTEM EVOLUTION

A Comparative Study of Modernisation and Replacement Initiation Factors

Irja Kankaanpää

Information Technology Research Institute, University of Jyväskylä, Mattilanniemi 2, Jyväskylä, Finland

Päivi Tiihonen

Jyväskylä University of Applied Science, Piippukatu 3, Jyväskylä, Finland

Jarmo J. Ahonen

Department of Computer Science, University of Kuopio, Kuopio, Finland

Jussi Koskinen

Department of Computer Science and Information Systems, University of Jyväskylä, Jyväskylä, Finland

Tero Tilus, Henna Sivula

Keywords: Legacy system evolution, evolution initiation, decision making, initiation factor.

Abstract: Decisions regarding information system evolution strategy become topical as the organisation’s information

systems age and start to approach the end of their life cycle. An interview study was conducted in order to

compare factors influencing modernisation and replacement initiation. System age, obsolete technology and

high operation or maintenance costs were identified as triggers for both modernisation and replacement

projects. The results show that the most prevalent individual reason for modernisation initiative is business

development. Common initiation factors for replacement projects were end of vendor support and system’s

inability to respond to organisation’s business needs.

1 INTRODUCTION

Continuous evolution is necessary in order to

maintain a system’s ability to respond to the

requirements of its environment (Lehman 1998).

When a system ages and conventional maintenance

faces increasing difficulties with system up-dates the

organisation often faces a ”legacy dilemma”

(Bennett 1995). A system with a long lifetime often

contains accumulated business critical data yet

preserving its functionality as such would require

extensive use of resources and face difficulties due

to obsolete technology (Bennett 1995).

There are three strategies to tackle a legacy

system: 1) continuing maintenance despite of

possible complications, 2) replacing the legacy

system with a new system, or 3) modernising it

(Bennett 1995, Seacord et al. 2003, p. 8-10).

Modernization and replacement are significant

economical investments with wide range of

organisational effects. Consequently, evolution

decisions are of great importance. In practice, the

280

Kankaanpää I., Tiihonen P., J. Ahonen J., Koskinen J., Tilus T. and Sivula H. (2007).

LEGACY SYSTEM EVOLUTION - A Comparative Study of Modernisation and Replacement Initiation Factors.

In Proceedings of the Ninth International Conference on Enterprise Information Systems - DISI, pages 280-287

DOI: 10.5220/0002378102800287

 SciTePress

assessment of potential evolution options is

challenging and decisions are often made informally

and largely based on intuition (Saarelainen et al.

2006).

The aim of this study is to compare replacement

and modernisation projects in order to provide an

understanding of the factors influencing evolution

initiation. A comparative study is important because

when the differences and similarities of evolution

types are known they can be taken into consideration

in evolution planning and decision processes. The

term evolution initiative refers to “a formally

established and organized effort to evolve a system”

(Bergey et al. 1997). Here, the term initiation factor

refers to the decision basis and the reasons that

trigger evolution activities.

2 EVOLUTION OPTIONS AND

DECISION MAKING

In the past, several decision frameworks and models

have been developed to systemise legacy system

management and decision making. These include

SABA – a decision model for legacy systems

(Bennett et al. 1999), Renaissance (Warren and

Ransom 2002), A decision framework for Legacy

System Management (De Lucia et al. 2001), and

Enterprise Framework for the Disciplined Evolution

of Legacy Systems (Bergey et al. 1997). All these

models emphasise the fact that legacy system

decisions should include consideration of a number

of factors and that the decision process should be

rather formal in order to assure the success of system

evolution. It has been proposed that the essential

elements influencing legacy system evolution

initiative are organisation, project, legacy system,

target system, systems engineering, software

engineering and technologies (Bergey et al. 1997).

Modernisation aims at improving the existing

system to correspond with the requirements of its

environment. It involves such radical modification

that it cannot be considered as regular maintenance

(Seacord et al. 2003, p. 9). Despite the significant

improvements, a considerable proportion of the

legacy system is conserved in the modernisation

(Seacord et al. 2003, p. 9). Aversano et al. (2004)

propose that prior to modernisation decision the

technical and the business value of the existing

system should be assessed. Koskinen et al. (2005)

list twenty decision criteria influencing

modernisation decision making, the ten most

important criteria being: system usability, end of

technological support, changes in business

processes, maintenance costs, system correctness,

system efficiency, expected remaining system

lifetime, size of required changes, application

domain expertise of maintainers, and delocalized

system logic.

Replacement refers to an activity as a result of

which a legacy system is replaced with another

system providing the same functionalities (Warren

and Ransom 2002). The replacing system can be

purchased as an off-the-shelf application (Seacord

2003, p. 10) or it can be a result of a redevelopment

project (Seacord 2003, p. 10, Bisbal et al. 1999).

Bandor (2006) suggests that software selection

process often focuses on the system functionality

while ignoring the intangible factors related to the

use and organisational influences of the system, and

system management and operation related risks.

3 RESEARCH METHODS

The goal of the study was to compare two evolution

types, i.e. modernisation and replacement, with

respect to evolution initiation triggers. Therefore, it

was relevant to collect empirical evidence on the

topic. A qualitative research approach was chosen

for it was desired to produce new information on the

topic and not only to verify the results of previous

research. For this purpose, semi-structured interview

was considered as the most suitable data collection

method (see Seaman 1999). Interview topics were

selected and questions formulated based on a

literature survey.

A set of pre-requirements were defined for

acceptable projects for the interviews. The selection

criteria were: 1) the project fitted either the

modernisation or the replacement definition

presented earlier, 2) the project was completed, and

3) a new or modernised system had been

implemented in the organisation by the time of the

interview. Also, an interview could relate to one

evolution project only.

3.1 Data Acquisition

Data was acquired with phone interviews. Randomly

picked organisations were contacted and inquired

whether there had been suitable modernisation or

replacement projects in the organisation’s past

(between years 2000 and 2005). If the response was

positive, a key person involved in the project was

contacted and requested an interview. The suitability

of projects was assessed with the key person before

LEGACY SYSTEM EVOLUTION – A Comparative Study of Modernisation and Replacement Initiation Factors

281

the interview. If there had been more than one

project that fitted the given criteria it was negotiated

with the key person which project would be best

suited to for discussion.

The person interviewed was most commonly the

person with the deepest involvement or knowledge

about the selected project. Prior to the interview, the

interviewees were sent a list of the topics that were

to be covered in the interview so that they could

refresh their memory on the project and check

documents or other data sources if necessary.

Data collection took place between autumn 2004

and autumn 2005. A total of 60 interviews were

gained but all of them could not be used in the

analysis for the following reasons: the project did

not meet with the pre-requirements; or the project

was a borderline case between the definitions of

modernisation and replacement. These interviews

were excluded from the analysis. The final material

sample consists of 29 interviews from which 14

concerned modernisation and 15 replacement

projects. Data was collected from 28 organisations,

including both private companies and public

organisations, and from 29 evolution projects within.

One person was interviewed per project. One

organisation yielded two suitable projects and thus

the number of projects (and interviews) is higher

than the number of organisations.

Most of the interviewees were upper or middle

level managers in data administration. The sample

also included IT development executives and other

IT personnel. The study focused on user

organisations and, consequently, system suppliers

were excluded.

3.2 Data Analysis

Interviews were recorded by the permission of the

interviewee and then transcribed. Data collection

was closed when the material was saturated, i.e. new

cases did not bring new information on the topic. In

data analysis, classification by theme, type, and

content breakdown were used. Before the analysis

the material was read through several times in order

to gain familiarity with it and to identify features

that describe the material in general. The material

was coded by theme in order to increase the fluency

of handling. After the theme classification, the

material was classified according to the frequency of

response types in order to increase the comparability

of the results.

3.3 Reliability of the Results

The reliability of the results was assured by using

two researchers to analyse the material. They were

not connected to the organisations and approached

the material from an outsider’s perspective. It was

noted for that the reliability of the results could be

weakened by the fact that the material was relatively

heterogeneous, i.e. it consisted of different types and

sizes of organisations and information systems.

However, the responses were fairly homogenous in

both evolution types. This indicates that the diversity

in target organisations and systems have not

undermined the reliability of the results.

The use of semi-structured interview provided a

possibility to acquire in-depth information on the

same topic from a variety of interviewees. The

issues and viewpoints the interviewees brought up in

addition to the pre-designed questions were relevant

to the topic and added depth to the material. The

reliability of the results could have been improved

with additional quantitative data collection.

4 RESULTS

In the following, the results on evolution initiation

comparison are reported. Citations

from the

interviews are included in the text in order to clarify

the results. The proportion of the appearance of each

factor with respect to the total amount of projects is

given in brackets (number of appearance / total

amount). Abbreviation “R” is used for replacement

and “M” for modernisation.

A total of 15 initiation factors were identified in

the study. Some of these appeared in both

modernisation and replacement projects, while some

were evolution type specific. Factors influencing

both modernisation and replacement initiation are

system age, obsolete technology and high operation

or maintenance costs. Identified replacement

specific initiation factors were end of vendor

support, need to unify disintegrated systems and

scattered data, need for a system compatible with

organisation’s other systems, software or hardware,

and to follow organisation’s IT strategy or the

prevailing IT trends within the industry.

Modernisation specific initiation triggers were the

desire to develop organisation’s business or business

processes, the requirements presented by a business

Citations have been translated from Finnish to English

by the author.

ICEIS 2007 - International Conference on Enterprise Information Systems

282

partner, responding to customer needs, legislation

changes, and competitive advantage.

Table 1 summarises the initiation reasons and the

number of their occurrence in replacement (column

“R”) and modernisation (column “M”) projects. The

number or occurrence presents the total number of

projects where a presented factor appeared.

Table 1: Reasons for evolution initiation.

Reason for evolution initiation R M

System age 9 5

Obsolete technology 3 4

High maintenance or operation costs 1 2

Maintenance difficult or not possible 1 3

End of vendor support 7 -

Incompliance with business needs 7 -

Disintegrated systems and scattered data 3 -

Incompatibility with other systems 3 -

Incompliance with IT strategy 2 -

IT trends in the field of industry 1 -

Business or business process development - 7

Customer needs - 2

Business partner requirements - 1

Change of legislation - 1

Competitive advance - 1

4.1 Replacement Initiation

The results suggest that replacement initiation is

triggered by ten factors. These factors and their

distribution throughout the replacement projects is

depicted in Table 2, where R# denotes replacement

project and x indicates the appearance of initiating

factor in a project. Column T

gives the total

number of initiation factors’ appearance across the

replacement projects.

Three factors appeared significantly more

frequently than the others: the system age (9/15), the

end of vendor support (7/15), and the system’s

inability to respond to company’s current business

needs (7/15). Often these three major reasons

appeared together or combined with other factors.

(R7): "The system was mainly from the year 1985

and then further developed. Its basic structure was

nearly 20 years old … operations had changed

significantly during that time and business

requirements had changed so much that it was

decided to renew the system.”

In almost half of the replacement projects (7/15),

the reason for acquiring a new system was the

legacy system supplier that had stopped or had

announced the termination of system updates,

maintenance or technical support. Typically, the

vendor had informed the customer that development

of the system would be discontinued. In R5, R11 and

R12, the existing system was not compatible with

other systems in its operational environment, with

new operating systems or up-dated hardware. In the

following, an interviewee (R5) describes a project

where both of these factors were present: ”… we had

a very old, about 10 years old, information system in

use and we were in a situation that we could not

really get updates for it and it was not compatible

with these new operating systems any more. So in

that situation we had to renew the system.”

In R4, R9 and R12, the legacy system consisted

of interoperating sub-systems and problems with

scattered data arose. The goal was to unify separate

systems by replacing them with a single ERP system

that would store all data in one location. An

interviewee (R4) concludes: “…when there are

various small systems and their maintenance costs

are fairly high we aimed at [getting them] within

one system. On the other hand we were thinking

about the transparency of customer data, and that

we were transferring the same data in so many

places, so that was one reason why we wanted to get

rid of those separate systems.”

In addition to old age and disintegrated systems, IT

trends in the field of industry encouraged system

replacement (R12): Well, it [system] was aged and it

consisted of separate systems and at some point we

should have changed it anyway … and what I have

heard is that those [companies] who used the same

systems, they have had quite a strong changing wave

going on.”

Organisation’s IT strategy guided the decision

making in two cases. In R12, the system was

completely rewritten in order to comply with new

architecture and operating system requirements

defined in the IT strategy. In R4, according to

organisation’s IT strategy all systems were to be

united and, hence, individual systems were

integrated into one organisation wide ERP system.

4.2 Modernisation Initiation

The results suggest that modernisation initiation is

influenced by nine factors. From these three types of

modernisation initiatives can be identified: business

development driven modernisation, legacy system

The largest individual modernisation trigger was the

desire to develop the organisation’s business or

business processes (7/14). It was typical that the

weaknesses of the legacy system were

acknowledged and action was taken in order to

improve related operations.

Modernisation activities were considered as a

natural part of organisation’s operations and they

took place in regular intervals (M11): “This project,

LEGACY SYSTEM EVOLUTION – A Comparative Study of Modernisation and Replacement Initiation Factors

283

we called it ‘the development of IT environment for

better business process support’. … the whole IT

environment and architecture was developed. It

supports business processes and business objectives

better. … one can say that it is an endless [process]

but now [in this project] the basic improvements

have been done.”

In one case, where modernisation was seen as

an enabler of organisational development, it was

anticipated to provide competitive advance. The

motive for modernisation was to be the first

organisation in the national markets to use new

technology (M12): ”Yes, it has reference value if we

are the first ones to do something.” In another case,

the prevailing IT trends within the industry were

regularly reviewed and assessed from business

development perspective (M5):

“I belong to a national data administration group

and we meet fairly regularly and discuss about the

state of the art of software development in our line

of business.”

In four modernisation projects the initiative

originated solely from business development

motives (M4, M12) or as response to external

pressure (M6, M14). In M6, legislation had changed

and required rapid changes in the system. In M14,

modernisation was initiated by the change request of

an important business partner and responding to it

was perceived as an opportunity for system

improvement to better respond to customer needs

and for business development.

The results indicate that legacy system related

factors form the largest group of modernisation

initiation reasons. Old age (6/14), obsolete

technology (5/14) and impossible maintenance

(3/14) were typically mentioned. Economical factors

accompanied obsolete technology. High operation

costs (2/14) were additional motivators. In extreme

cases, old age (M3, M9) or obsolete technology

(M7) was the only reason for modernisation.

The third identified modernisation initiation type

includes both technical and business reasons: system

was old and did not serve the user organisation as

well as in the past (M5), system was old and did not

respond to new customer needs (M1), maintenance

was not possible and organisation wanted to develop

its business operations by modernising the existing

system (M11), or maintenance was nearly

impossible due to obsolete technology and system

Table 2: Reasons for replacement initiation.

Initiation factors R1 R2 R3 R4 R5 R6 R7 R8 R9 R10 R11 R12 R13 R14 R15 T

System age x x x x x x x x x 9

Incompliance with business needs x x x x x x x 7

End of vendor support x x x x x x x 7

Obsolete technology x x x 3

Incompatibility with other systems x x x 3

Disintegrated systems and scattered data x x x 3

Incompliance with IT strategy x x 2

IT trends in the field of industry x 1

Difficult maintenance x 1

High maintenance costs x 1

Table 3: Reasons for modernisation initiation.

Initiating factor M1 M2 M3 M4 M5 M6 M7 M8 M9 M10 M11 M12 M13 M14 T

Business or business process

development

x x x x x x x 7

System age x x x x 4

Obsolete technology x x x x 4

Maintenance not possible x x x 3

High operation costs x x 2

Customer needs x x 2

Business partner requirements x 1

Change of legislation x 1

Competitive advance x 1

ICEIS 2007 - International Conference on Enterprise Information Systems

284

Table 4: Comparison of evolution initiation criteria.

No. Modernization decision criteria (D)

(Koskinen et al. 2005)

Modernisation initiation factors

)

Replacement initiation factors

)

1. System usability Business development System age

2. End of technological support System age End of vendor support

3. Changes in business processes Obsolete technology Incompliance with business needs

4. Maintenance costs Maintenance not possible Obsolete technology

5. System correctness High maintenance costs Disintegrated systems and scattered

data

6. System efficiency Customer needs Incompatibility with other systems

7. Expected remaining system lifetime Business partner requirements Incompliance with IT strategy

8. Size of required changes Change of legislation IT trends in field of industry

9. Application domain expertise Competitive advance High maintenance costs

10. Delocalized system logic Maintenance difficult

… …

15. System age

16. Changes in business environment

was modernised in order to better supported business

processes and objectives (M13).

The distribution of initiating factors throughout the

modernisation projects is depicted in Table 3, where

M# denotes modernisation project and x indicates

the appearance of a factor in the left hand side

column in a project. Column T

gives the total

number of initiation factors across the modernisation

projects.

5 DISCUSSION

The reasons behind replacement and modernisation

initiatives were noticeably similar but varied in

nuances. Legacy system related factors proposed in

earlier studies (see e.g. Aversano et al. 2004, Bergey

et al. 1997), i.e. system age, obsolete technology, or

difficult and costly maintenance, influenced

evolution initiation in both project types.

A separating factor concerning evolution

strategy selection was the degree of evolution

initiative anticipation. Organisations that decided to

modernise their legacy system wanted to develop

their business processes. They were, in general,

aware of IT trends in their business line and

launched modernisation in order to develop their

business. The interviewees were not directly asked

about their organisation’s business perspective, yet it

was mentioned as an important factor in half of the

interviews. Thus, strategic business planning as

management activity (see Bergey et al. 1997) was a

central element. On the contrary, interviewees from

organisations that had chosen replacement as an

evolution strategy did not express the desire for

business development. Instead, replacement was a

reaction to the vendor’s announcement of ending

system support or to the system’s inability to support

organisation’s business operations. This would

suggest that a system management related risk

appeared real (see Bandor 2006) and caused action.

It is proposed that modernisation took place as a

result of proactive situation assessment while

replacement was a reaction to a change in

organisation’s internal or external environment. The

reasons behind this phenomenon require further

research.

The results show that legacy system replacement

and modernisation differ from each other with

respect to vendor influence on evolution initiation.

Nearly half of the replacement projects were

initiated because the system supplier had given up

system development or support. However, in

modernisation projects, vendor’s actions did not

influence evolution initiation. It is concluded that

organisations that decide to replace legacy systems

are more dependent on system suppliers than

organisations that choose modernisation.

When compared to the earlier findings, it can be

noticed that there is clear divergence between

initiation factors although similarities exist, too.

Table 4 lists the decision criteria reported by

Koskinen et al. (2005), which are in the text denoted

with D, and the findings of this study, denoted with

(modernisation initiation factors) and R

(replacement initiation factors) in a descending order

of importance. An interesting finding was that

system usability, which was listed as the most

LEGACY SYSTEM EVOLUTION – A Comparative Study of Modernisation and Replacement Initiation Factors

285

important modernisation decision criteria, did not

occur among the evolution initiation factors of found

in this study. An explanation for this phenomenon

was not found and would require further research.

Another observation is that the modernisation

decision criteria reported by Koskinen et al. (2005)

correspond to replacement initiation factors better

than modernisation triggers found in this study.

Correlation for all but one replacement initiation

factor, i.e. end of vendor support (R

2), exists: R

1 = D15, R

3 and R

7 = D3, R

4 = D2, R

5 and

6 = D5, R

8 = D16, R

9 = D4, and R

10 = D7.

5 and R

6 correspond to system efficiency (D6)

because they led to inefficient system use in studied

organisations. It should be noted that D2 refers to

general technology changes and, hence, it should not

be mixed with R

Six of the identified modernisation initiation

factors match with the previously presented criteria.

The following correlations exist: D2 = M

3, D4 =

5, D7 = M

4, D15 = M

2, and D16 = M

7. New factors, not appearing in the list by

Koskinen et al. (2005), are business or business

process development objective (M

1), end of

vendor support (R

2), and competitive advance

10).

The amount of factors influencing evolution

initiation decision is significantly smaller than given

before (see Koskinen et al. 2005). The results

strengthen the previous findings of the importance of

engaging both business and technical aspects in

evolution decision making (see Aversano et al. 2004,

Bergey et al. 1997).

6 SUMMARY

The aim of this study was to compare replacement

and modernisation projects with respect to the

reasons that initiate evolution activities in order to

provide evidence on their differences and

similarities to support evolution planning and

decision processes. The differences and similarities

were mapped with an empirical study where data

administration managers, IT development executives

and other IT personnel from 29 evolution projects

were interviewed.

The results confirm the previously suggested

modernisation criteria but challenge the order of

their importance. The findings question the

importance of system usability, previously claimed

to the most important modernisation decision

criteria. New factors complementing earlier findings

are business or business process development

objective, end of vendor support, and gaining

competitive advantage.

System age, obsolete technology and high

operation or maintenance costs were identified as

triggers in both evolution types. The most common

initiation factors in replacement projects were

system age, end of vendor support, and system’s

incompliance with the organisation’s business needs.

The most common reasons for modernisation were

the desire to develop organisation’s business or

business processes, system age, and obsolete

technology.

ACKNOWLEDGEMENTS

This work has been carried out in ITRI (Information

Technology Research Institute) and financially

supported by TEKES (Finnish Funding Agency for

Technology and Innovation).

REFERENCES

Aversano, L., Esposito, R., Mallardo, T. and Tortorella,

M., 2004. Supporting Decisions on the Adoption of

Re-engineering Technologies. In CSMR’04, 8ht

European Conference on Software Maintenance and

Reengineering. IEEE Computer Society, 95-104

Bandor, M. S., 2006. Quantitative Methods for Software

Selection and Evaluation, Technical Note, CMU/SEI-

2006-TN-026. Retrieved November 15, 2006 from

Carnegie Mellon University, Software Engineering

Institute Web site: http://www.sei.cmu.edu/

publications/documents/06.reports/06tn026.html

Bennett, K., 1995. Legacy Systems: Coping with Success.

IEEE Software, 12 (1), 19-23.

Bennett, K., Ramage, M. and Munro, M., 1999. Decision

Model for Legacy Systems. IEE Proceedings -

Software 146 (3), 153-159.

Bergey, J. K., Northrop, L. M. and Smith, D. B., 1997.

Enterprise Framework for the Disciplined Evolution

of Legacy Systems, Technical Report, CMU/SEI-97-

TR-007, ESC-TR-97-007, Retrieved November 15,

2006 from Carnegie Mellon University, Software

Engineering Institute Web site: http://www.sei.cmu.

edu/publications/documents/97.reports/97tr007/97tr00

7abstract.html

Bisbal, J., Lawless, D., Wu, B. and Grimson J., 1999.

Legacy Information Systems: Issues and Directions.

IEEE Software, 16 (5), 103-111.

De Lucia, A., Fasolino, A.R. and Pompella, E., 2001. A

Decisional Framework for Legacy System

Management, In ICSM’01, International Conference

on Software Maintenance, IEEE, 642-651.

Koskinen, J., Ahonen, J.J., Sivula, H., Tilus, T., Lintinen,

H. and Kankaanpää, I., 2005. Software Modernization

ICEIS 2007 - International Conference on Enterprise Information Systems

286

Decision Criteria: An Empirical Study, In CSMR’05,

9th European Conference on Software Maintenance

and Reengineering. IEEE Computer Society, 324-331.

Lehman, M. M., Perry, D. E. and Ramil, J. F., 1998.

Implications of Evolution Metrics on Software

Maintenance. In International Conference on

Software. Maintenance. IEEE Computer Society, 208-

217.

Saarelainen M., Ahonen, J.J., Lintinen, H., Koskinen, J.,

Kankaanpää, I., Sivula, H., Juutilainen, P. and Tilus,

T., 2006. Software Modernization and Replacement

Decision Making in Industry: A Qualitative Study. In

EASE’06, 10th International Conference on

Evaluation and Assessment in Software Engineering.

Kitchenham, B.A., Brereton, P., Turner, M., and

Charters, S. (eds.), Retrieved December 5, 2006 from

BCS eWIC database.

Seacord, R. C., Plakosh, D. and Lewis G. A., 2003.

Modernizing Legacy Systems – Software

Technologies, Engineering Processes and Business

Practices. Addison-Wesley, Boston, MA.

Seaman, C. B., 1999. Qualitative Methods in Empirical

Studies of Software Engineering, IEEE Transactions

on Software Engineering, 25 (4), 557-572.

Warren, I. and Ransom, J., 2002. Renaissance: A Method

to Support Software System Evolution. In

COMPSAC’02, 26

Annual International Computer

Software and Applications Conference. IEEE, 415-

420.

LEGACY SYSTEM EVOLUTION – A Comparative Study of Modernisation and Replacement Initiation Factors

287