=Paper=
{{Paper
|id=Vol-2443/paper09
|storemode=property
|title=Predicting Next Wave of Digitalization: Towards a Theory of Evolution of Enterprise Applications
|pdfUrl=https://ceur-ws.org/Vol-2443/paper09.pdf
|volume=Vol-2443
|authors=Janis Grabis
|dblpUrl=https://dblp.org/rec/conf/bir/Grabis19
}}
==Predicting Next Wave of Digitalization: Towards a Theory of Evolution of Enterprise Applications==
https://ceur-ws.org/Vol-2443/paper09.pdf
Predicting Next Wave of Digitalization: Towards a
Theory of Evolution of Enterprise Applications
Jānis Grabis
Riga Technical University, Institute of Information Technology
Kalku 1, Riga, LV-1658, Latvia
grabis@rtu.lv
Abstract. There is a continuous strive for digitalization of business process exe-
cution and different types of enterprise applications are used to implement busi-
ness processes. The paper identifies types of enterprise applications, namely,
functional applications, ERP systems, workflow management systems, service-
oriented architecture applications and SaaS and analyzes suitability of these ap-
plications for fulfilling business process execution requirements. This analysis
yields a hypothesis that enterprise applications evolve cyclically and the cycles
are determined by an interplay between standardization and customization
tendencies in development of enterprise applications. This observation allows to
structure existing developments in enterprise computing and to reason about fu-
ture developments in digitalization of business processes.
Keywords: Business processes, digitalization, enterprise applications, evolution
1 Introduction
Companies make products and deliver services by performing business processes.
There is large variety of business processes and they are interrelated. Enterprise appli-
cations are used to execute these business processes. Over the course of time, new tech-
nologies are becoming available to implement these enterprise applications. Occasion-
ally, these new technologies cause a significant overhaul of enterprise applications by
introducing new paradigms, platforms, capabilities and functionality. There is a contin-
uous evolution of enterprise applications and several generations of applications can be
identified. Comprehensive digitalization of enterprise business processes is the latest
wave of adopting information technologies by businesses.
Business process improvement is a continuous activity and digitalization technolo-
gies play a major role in achieving the whole potential of process improvement [1].
Enterprise Resource Planning (ERP) systems were one of the first integrated solutions
for running enterprise business processes [2]. Workflow Management (WfM) systems
at one point arose as an alternative solution providing explicit representation of busi-
ness processes [3]. Service-oriented architecture (SOA) applications distinguished
themselves with increased flexibility and cross-enterprise integration [4]. New oppor-
tunities for simplifying application management and execution of data intensive busi-
ness processes were introduced by Software a as Service (SaaS) applications [5, 6]. At
the height of their popularity, many of these applications were perceived as
Copyright © 2019 for this paper by its authors. Use permitted under Creative Commons
License Attribution 4.0 International (CC BY 4.0)
98
replacements of other types of applications although further developments showed that
every type has its specific strengths and weaknesses resulting in complementary appli-
cations. Obviously, areas of emphasis in enterprise applications have been changing
over time. Shankararaman et al. describe four waves of evolution of enterprise systems
starting with systems focusing on task automation to integrated systems and agile col-
laborative systems [7]. ERP systems themselves have undergone multiple stages of
evolution [8]. Business process and enterprise integration has been acknowledged as
one of the main driving forces in evolution enterprise information systems [9]. How-
ever, these investigations mainly focus on one type of enterprise applications.
The objective of this paper is to propose a framework for describing evolution of
enterprise applications according to fulfillment of process execution requirements. The
process execution requirements are based on typical characteristics of business pro-
cesses and their performance attributes. An ability of enterprise applications to meet
these requirements is analyzed using general characteristics and features of these appli-
cations as the evaluation framework. The results of the analysis are summarized as a
hypothesis about evolution cycles of enterprise applications. The paper is not intended
to provide comprehensive review of enterprise applications and their properties rather
it reflects upon experiences and observations accumulated over more than fifteen years
of academic work in the area.
There are two main contributions of the paper: 1) summary on evaluation of types
of enterprise applications according to their business process execution performance;
and 2) formulation of the hypothesis about evolution cycles of enterprise applications.
The rest of the paper is organized as follows. Section 2 describes general require-
ments towards enterprise applications and defines performance attributes for evaluation
of these applications. Section 3 provides an overview of types of enterprise applica-
tions. The hypothesis on cyclical evolution of enterprise applications is formulated in
Section 4. Section 5 concludes.
2 Requirements
Business process execution needs pose requirements towards enterprise applications.
These requirements are affected by typical business process characteristics:
• Variety – companies have a large number of different processes ranging from pro-
curement to sales and finance to human resources and every process has its own
specific features;
• Multiple stakeholders both internal and external – different stakeholders are in-
volved in business processes and they should be able to collaborate efficiently;
• Interdependencies – business processes are intertwined and depend on each other.
That requires information exchange and coordination among the business processes;
• Large volume – companies have many employees and partners executing many pro-
cess instances creating large computational load;
• Dynamic – business processes change frequently and enterprise applications need to
be updated to reflect these changes.
99
The ultimate goal of using enterprise applications in business process execution is
improvement of business process performance. The business process performance can
be evaluated along four dimensions [10]:
• Cost
• Time
• Quality
• Flexibility
The performance attributes can be evaluated using variety of specific measures. In
this paper their interpretation is considered in relation to enterprise applications.
The cost attribute is influenced by several aspects, notably, development and execu-
tion costs. If an application provides specialized means for development without much
reliance on proprietary technologies then development cost is lower. If execution does
not incur significant overhead and does not require large computational resources then
execution cost is lower.
The time attribute characterizes ease and speed of business process execution from
the user’s perspective. If an application provides task specific means for efficient pro-
cess execution (e.g., data input is optimized in call center customer service process) it
contributes to improving the time attribute. On the other hand if an application is built
for every purpose and intent and provides generic support to business process execu-
tion, its efficiency is not that high.
From the business process execution perspective, quality concerns providing the
right information for performing business tasks correctly. The level of integration
among business processes is a significant factor affecting this ability. If an application
is able to deliver information to users in a timely and suitable manner then it contributes
to improving the quality attribute.
Flexibility characterizes an ability to deal with heterogeneity and changing nature of
business processes. If an application can be adapted to serve various types of processes
and can be modified when necessary then its flexibility is high.
3 Types of Enterprise Applications
Five types of enterprise applications, namely, functional applications, ERP systems,
WfM systems, SOA applications and SaaS are considered in this paper. This section
describes general characteristics of enterprise applications and peculiarities associated
with every type.
3.1 General Characteristics
The enterprise applications are applications used to execute business processes. Their
common features are:
• Support for large user base – many geographically distributed users access applica-
tions concurrently;
100
• Support for data and process integration – users are able to share data and pass exe-
cution control across multiple processes;
• Computationally intensive operations – applications implement complex decision-
making logics and provide information for data-driven processes;
• Reuse – many companies and processes require similar functionality, which is re-
used as often as possible at various levels of granularity;
• Modifiability – to serve unique and changing requirements enterprise applications
provides means for customizing business processes.
The enterprise applications are very large and highly complex applications. Their
development is time and resource consuming. Therefore, some sort of reuse features
prominently to reduce this development complexity. That ranges from reuse of individ-
ual components to whole package and enterprise applications are frequently provided
by their vendors as commercial-of-the-shelf packages regardless of the type of delivery
and licensing mode. The packaged applications often encompass business process ex-
ecution knowledge and some of the processes have become commoditized or standard-
ized. In order to support custom requirements in the packaged applications, various
modification options are provided [11]. These range from high level code-free config-
uration to low-level modification by coding.
Utilization of standard reusable features reduces costs while modifications are aimed
at improving business process execution efficiency. Finding a right balance between
usage of standard features and customization is an ongoing discussion [12, 13]. This
interplay between using standard features and customization can be generalized to other
features as well (Table 1). Customization leaning enterprise applications implement
custom processes, use proprietary technologies, are developed according to specialized
designs (e.g., technical design, user interface) or rely on decentralization (e.g., data
sharing, computations). Standardization leaning enterprise applications provide stand-
ard processes, use open technologies, have unified design and are centralized to some
extent.
Table 1. Customized and standardized features of enterprise applications.
Customized features Standardization features
Custom process Standard process
Proprietary technologies Open technologies
Specialized design Unified design
Decentralization Centralization
3.2 Functional Applications and ERP Systems
Functional applications are intended for supporting a limited set of business process in
a certain functional area of the company [14]. Their typical characteristics are: 1) cus-
tom-built; 2) limited integration; and 3) usage of myriad of technologies in implemen-
tation. The functional applications provide basic data processing and process tracking
functionality though they do not support majority of the requirements towards enter-
prise applications.
101
ERP systems are large packaged applications aimed at automation of majority of
enterprise business processes. They consist of multiple functional modules joining to-
gether related business processes in one area. ERP systems typically have three tier
client-server architecture where centralized database serves as means for company-
wide business process integration. The best practice processes are implemented in ERP
systems and all modules typically have the same design. Although processes can be
configured to the needs of a particular enterprise, they are hard-coded and somewhat
obscure to end-users. The ERP systems are data-oriented applications. There are tools
for customizing ERP systems. Some of these tools are special purpose tools while usage
of general purpose development tools gradually increases.
3.3 Workflow Management Systems
The Workflow Management (WfM) systems introduced explicit representation of busi-
ness processes are used to develop process-oriented applications, which execute a se-
quence of tasks. They allow definition of custom process logics and execution of these
custom processes using specialized execution engine. Any task could be included in the
custom process definition though WfM systems do not provide its business logics out-
of-the-box and often tasks are limited to handling data input/output operations.
Vendors of WfM initially developed their own methods and tools for development
and execution although BPMN recently is widely used to define processes. WfM sys-
tems were typically built according to the WfM systems reference architecture. WfM
systems do not necessarily relay on centralized data storage and data can be passed
directly from one task to another. That introduces a level of decentralization in WfM
based applications.
3.4 SOA Applications
SOA applications are developed by composing services provided by their providers.
That potentially provides a high degree of reusability and possibility to modify appli-
cations by replacing services. Web services are the most often used type of services.
The web services use open standards making applications widely accessible and porta-
ble. That leads to suitability of service-orientation for inter-enterprise integration. Exe-
cutable business process modeling languages provide possibilities for model-driven or-
chestration of the services, which provide reusable business logics.
SOA is used to develop and execute SOA applications and the enterprise service bus
is often used to ensure communications among services. Despite services being distrib-
uted across multiple hosts, the service bus introduced some level of centralization in
the system.
The main drawback of SOA applications is execution overhead caused by excessive
communication among services. That also increases application execution complexity
and requires allocation of significant computational resources.
102
3.5 SaaS
SaaS similarly as SOA applications follow the servitization paradigm. In this paper, the
SOA applications are strongly associated with SOA as the development and execution
platform while SaaS is the end-user facing cloud computing service model. In the case
of SaaS, the whole enterprise application is provided over the Internet in a service mode
[15]. From the consumer perspective, SaaS provides immediate access to required func-
tionality while control over running and modifying the application is limited. From the
provider perspective, SaaS provides cost saving opportunities chiefly due to multi-ten-
ancy. These applications are deployed in the cloud environment inheriting cloud ena-
bled properties such as scalability to deal with computationally intensive operations.
SaaS applications can be built following variety of architectural styles. Recently, the
microservice architecture [16] has become one of the dominant solutions. It furthers
service-orientation by allowing each service to be developed and deployed inde-
pendently using its own technology stack. The services communicate among each other
using their API. Although web technologies based REST API are frequently used, other
types of API are also available and their usage increases.
Similarly, to SOA applications there is an execution overhead amplified by decen-
tralized integration links.
4 Evolution Cycles
Section 3 shows that consecutive generations of enterprise application all have their
strengths and limitations. Table 2 summarizes evaluation of types of enterprise appli-
cations according to the performance attributes. The main limitation of functional ap-
plications is their isolationary nature causing costs increases and limited integration
while proprietary technologies make modification difficult if there is lack of skills. ERP
systems are built as generic applications what saves costs though might case inefficien-
cies for execution of highly specific processes. The usage of standard processes brings
benefits or reuse but limits company’s ability to respond to unexpected user require-
ments and switching vendors is complicated. WfM systems extend internal integration
to multiple applications. Although there are standards for making workflows compati-
ble and portable, only simple administrative workflows are reused and implementation
of business logics requires costly custom development. In the case of SOA applications,
rightsized services could be used efficiently to define applications suitable for specific
situations. Timewise, user interactions are often provided using universal web client
technologies, which despite their increasing efficiency are still not as efficient as native
client technologies. In the case of SaaS, there are often different incentives for service
providers and consumers. This paper mainly focuses on consumer benefits and assumes
that providers’ benefits are transferred to consumers. SaaS provides a complete busi-
ness process development and execution environment. Companies can start-using SaaS
quickly when a new business need arises. SaaS relies on web technologies extensively
though there is significant share of proprietary technologies as well. The consumers do
not have a full control of application and some specific requirements cannot imple-
mented.
103
Table 2. Evaluation of types of enterprise applications according to business process perfor-
mance attributes
Type Cost Time Quality Flexibility
Functional Little reuse (-) Possibly task Little integra- Proprietary
specific (+) tion (-) technologies
(-)
ERP Package level Generic (-) Internal inte- Standard pro-
reuse (+) gration (+) cesses (-)
Real-time pro- Proprietary
cessing (+) technologies
(-)
WfM Limited reuse Task specific Internal inte- Custom pro-
(-) (+) gration (+) cess develop-
ment (+)
Proprietary
technologies
(-)
SOA applica- Service level Universal (-) External inte- Open technol-
tions reuse (+) gration (+) ogies (+)
Execution
overhead (-)
SaaS API and Full- Highly task High-perfor- Quick ramp-
stack reuse (+) specific (+) mance pro- up (+)
Execution cessing (+) Modification
overhead (-) Variety of limitations (-)
consumers (+) Proprietary
technologies
(-)
Types of enterprise applications at least partially have evolved in parallel though the
interest in these types has developed chronologically as shown in Fig. 1. The first cycle
of evolution concerns functional applications. They were built according to case spe-
cific enterprise requirements using proprietary technologies what manifests the custom-
ization approach. The functional applications provided the base functionality for run-
ning enterprise applications. However, building custom solutions for every case is
costly and integration among individual applications was weak. The ERP systems in-
troduced standardized or best practice business processes integrated using the common
database. The costs were shared among users of packaged applications. The ERP sys-
tems can be perceived as a move towards standardized solutions. The standard pro-
cesses with some configuration served well for commodity processes though the one-
fit-all approach was not suitable for value-added processes. The WfM systems allowed
to specify custom business processes at the cost of limited reuse of business logics.
Usage of proprietary technologies and custom processes allow to attribute the WfM
systems to the customization approach. Service-orientation came along to further de-
velopment of custom processes and brought in reuse of business logics by means of
web services. It also adopted web standards to enable integration among enterprises.
104
This way SOA belongs to the standardization wave. The major shortcoming of SOA
was additional application execution overhead. SaaS applications are mainly intro-
duced to reduce cost and shorten application ramp-up time. The microservice architec-
ture allows building very specialized, for instance, high performance data analysis, ser-
vices accessible through multitude of channels. That and frequent usage of proprietary
technologies for cloud services suggests that this wave belongs to the customization
wave.
Customization
Functional WfM SaaS
Service- Next
ERP
oriented wave
Standardization
Fig. 1. Evolution cycle of enterprise applications
It can be observed that standardization or customization traits may manifest differ-
ently in every phase. Additionally, there is no claim that either customization or stand-
ardization is a preferred way of enterprise application delivery. Different types of ap-
plications are used simultaneously and jointly, for example, many ERP systems as well
as SaaS include workflow development capabilities. However, observing this pattern
of oscillation it could be hypothesized that:
Evolution of enterprise applications follows a cyclical pattern fluctuating from
customization to standardization.
If the hypothesis is true then the next wave of enterprise application evolution will
focus of standardization. The standardization aspects could be related to using decen-
tralized technologies to achieve benefits of centralization (in a similar fashion as block-
chains), promoting openness (e.g., open data usage and data markets) and seamless
integration of various architecture styles.
5 Conclusion
The paper has demonstrated that evolution of enterprise applications has responded to
challenges faced by companies in business process execution. The evolution has been
strongly affected by customization and standardization tendencies in development of
enterprise applications. Similar patterns can be observed in other areas of information
technology as well, for instance, fluctuation between centralized and decentralized
computing in IT management and emergence of two-speed IT in product development.
It is important to note that evolution of enterprise applications is not a linear, sequen-
tial process and various types of enterprise applications are complementary. Similarly,
the evaluation of enterprise applications according to the performance attributes focuses
only on the main tendencies and there are overlapping, contradictions and alternative
105
interpretations. Enterprise applications evolve along multiple perspectives (develop-
ment, usage, customization, etc.) and these perspectives are often difficult to separate
and the overall picture would grow overly complex.
The main limitation of the paper is that hypothesis is based on limited number of
observations and there are no easy means to prove it. One of possibilities is to look at
evolution of individual enterprise applications to trace possible alterations between cus-
tomization and standardization. An indirect evidence shows that indeed in practice the
development approaches fluctuate in time. Evaluation by means of extended literature
analysis or expert interviews also would be prone to judgmental interpretation of ob-
servations and would have limited practical value.
References
1. Denner, M. Püschel, L.C., Röglinger, M. How to Exploit the Digitalization Potential of Busi-
ness Processes, Business and Information Systems Engineering, 60, 4, 331-349 (2018)
2. Shehab, E.M., M.W. Sharp, L. Supramaniam and T.A. Spedding. Enterprise resource plan-
ning, Business Process Management Journal, Vol. 10 No. 4, pp. 359-386 (2004)
3. Cardoso, J., Bostrom, R. P., Sheth, A. P. Workflow Management Systems and ERP Systems:
Differences, Commonalities, and Applications. Information Technology and Management,
5 (3-4), 319-338 (2004)
4. Papazoglou, M.P., Traverso, P., Dustdar, S., Leymann, F. Service-oriented computing: State
of the art and research challenges, Computer, vol. 40, no. 11, pp. 38-45 (2007)
5. Benlian, A., Hess, T. Opportunities and risks of software-as-a-service: Findings from a sur-
vey of IT executives. Decision Support Systems Volume 52, Issue 1, 232-246 (2011)
6. Hofmann, P. ERP is dead, long live ERP, IEEE Internet Computing, vol. 12, no. 4, pp. 84-
88 (2008)
7. Shankararaman, V., Eng Kit, L. Dale, S. Integrating enterprise system's 3rd wave into IS
curriculum, 18th Americas Conference on Information Systems 2012, AMCIS 2012, pp.
1887 (2012)
8. Chung, S.H., Snyder, C.S. ERP adoption: a technological evolution approach, International
Journal of Agile Management Systems, Vol. 2 No. 1, pp. 24-32 (2000)
9. Romero, D., Vernadat, F. Enterprise information systems state of the art: Past, present and
future trends Computers in Industry, Volume 79, 3-13 (2016)
10. Dumas, M., La Rosa, M., Mendling, J., Reijers, H. A. Fundamentals of Business Process
Management, Springer (2018)
11. Uppström, E., CM Lönn, M Hoffsten, J Thorström. New implications for customization of
ERP systems. 48th Hawaii International Conference on System Sciences, 4220-4229 (2015)
12. Parthasarathy, S., Sharma, S. Efficiency analysis of ERP packages - A customization per-
spective. Computers in Industry, vol. 82, pp. 19-27 (2016)
13. Grabis, J. Optimization of Gaps Resolution Strategy in Implementation of ERP Systems.
ICEIS (1), 84-92 (2019)
14. Laudon, J., Laudon, K.C. Essentials of Business Information Systems (7th Edition). Pren-
tice-Hall, Inc. Upper Saddle River, NJ, USA (2006)
15. Alex Peng, G.C., Gala, C. Cloud ERP: A new dilemma to modern organisations?, Journal
of Computer Information Systems, vol. 54, no. 4, pp. 22-30 (2014)
16. Dragoni N. et al. Microservices: Yesterday, Today, and Tomorrow. In: Mazzara M., Meyer
B. eds. Present and Ulterior Software Engineering. Springer (2017)
106