Charting the Landscape of Enterprise Architecture Complexity ...

4 downloads 2254 Views 166KB Size Report
complexity of EA will benefit interactions between business processes and IT. Also, it will ... in some different but related domains, such as Software. Engineering ...
Charting the Landscape of Enterprise Architecture Complexity Cybernetics A systematic literature analysis Jiong Fu, Aimin Luo, Xueshan Luo and Junxian Liu

Abstract— Complexity Cybernetics relating to Enterprise Architecture (EA) is a hot topic recently. Although many researchers, organizations made much progress in the past years, common understanding in this domain is still limited. In this paper, a systematic literature analysis on complexity cybernetics in relation to EA was presented, in order to chart a landscape of state-of-the-art and to summarize some suggestions for future research. By analyzing the contexts of the 33 papers collected by the review method, different classifications were conducted. Besides, an extensive analysis, including distribution of papers over time, regional distribution, type of publication, and author-group distribution, was studied. In the end, some discussions and suggestions were given. Our work contributes to getting a better understanding towards the domain and to building a new groundwork future research.

I. I NTRODUCTION OWADAYS, relationships between business processes and IT systems are becoming more complicated and disordered, since both business processes and IT systems are various and complex. As a result, while the gap between business processes and IT gets bigger, it becomes difficult for Enterprise Architecture (EA) application to succeed. EA evolution and EA integration get relatively more complex as well. The original reason of these problems is that: EA complexity often gets out of control. Recently, complexity cybernetics of EA has been becoming quite hot. Many researchers, organizations conducted studies around the topic in the past years. Good cybernetic complexity of EA will benefit interactions between business processes and IT. Also, it will help improve the success rate of EA application, and help ensure the quality of evolution and integration of EA. Unfortunately, no common understanding of EA complexity cybernetics is developed, and this domain remains young and immature. This topic needs to be better defined and refocused. In order to move ahead, to get a better understanding of the domain and to build a new groundwork future research, this paper presents a systematic literature review of complexity cybernetics of Enterprise Architecture.

N

Manuscript received January 10, 2016. Jiong Fu, Aimin Luo, Xueshan Luo and Junxian Liu are from Science and Technology on Information Systems Engineering Laboratory, National University of Defense Technology, Changsha, Hunan, P. R. China (email: {jiongfu, amluo, xsluo}@nudt.edu.cn, [email protected]). This work was supported by the National Natural Science Foundation of China (71571189).

We aim to achieve the two goals: (I) Charting a landscape of state-of-the-art relating to complexity cybernetics of EA. (II) Providing some suggestion for future research. This paper is organized as follows. Section 2 presents a definition of EA complexity and related studies. Section 3 discusses our review method. Section 4 details the concrete research step-by-step. Some discussions as well as our contributions are provided in Section 5. Section 6 concludes the paper by noting future works. II. R ELATED W ORK Complexity science was proposed for the first time in 1960’s, but it made little progress until 1980’s, during which period John Holland, Stuart Kauffmann together with some other scientists made great contribution to the domain. After that, many new and great achievements emerged in some different but related domains, such as Software Engineering, Computing Science, Algorithm and Complex Network. As to the domain of EA, what is the complexity? The definition is not unified. However, two parallel threads of understandings are commonly available: (I) EA complexity is a kind of metric of all nodes and relationships in an Enterprise Architecture [1]; (II) the other one is a kind of metric of all flow paths in an Enterprise Architecture [2]. In the past years, some academic surveys on EA complexity are conducted. For example, A.W. Schneider [3] proposed a framework with simple notation of complexity for EA cybernetics by identifying eight aspects of complexity and by categorizing them (the eight aspects of complexity) into four independent categories. Recently, Schneider [4] focused on Application Landscape (AL) complexity cybernetics by analyzing the complexity metrics from the results of a literature survey. In some other surveys, EA complexity are connected with some other domains, e.g., M. Maurer [5] classified approaches towards complexity cybernetics in systems engineering. Similarly, P.N. Lowe [6] provided an introduction to relationships between (I) System of Systems (SoS), (II) complexity, and (III) modeling and simulation. However, a whole landscape or big picture of EA complexity cybernetics is still unavailable, while further improvement sounds necessary.

III. R EVIEW M ETHOD A systematic literature review has its own method framework and basic guidelines. Our literature review method is based on the guidelines of [7] and [8]. As shown in Figure 1, the adopted process of systematic literature review method includes 8 steps: • Purpose of the literature review; • Protocol and training; • Searching for the literature; • Practical screen; • Quality appraisal; • Data extraction; • Synthesis of studies; • and Writing the review. The rest of this section is organized as follows. First, Subsection A introduces our research questions. Then, Subsection B describes research tasks based on our research questions. Finally, Subsection C describes the detailed steps of search strategy Purpose of the literature review

Protocol and training

Searching for the literature

analyzing the distribution of papers, as to published time, their regional distribution, type of publication, and authorgroups distribution. To answer RQ3, we should analyze the weakness regarding the state-of-the-art of EA complexity cybernetics, and summarize the literature review with some directions of future research. Research Questions

Research Tasks Background

RQ1: How to classify the current research on EA complexity cybernetics?

Model

RQ2: What is the current state of EA complexity cybernetics?

RQ3: What can we learn for the current state to improve the future research?

Fig. 2.

Method

Distribution over time

Regional distribution

Type of publication

Authors groups distribution

Current state

Weakness

Direction

Research questions and corresponding research tasks

C. Search Strategy Data extraction

Quality appraisal

Synthesis of studies

Writing the review

Fig. 1.

Practical screen

8-step systematic literature review method, by Okoli[7]

The search strategy directly shapes the quality of finding papers. In our review, the search process includes six parts, shown in Figure 3, including: Define Review Protocol, Select Search Engines, Define Keywords, Keyword-based Search, Pick Relevant Papers and Collect Papers. Define Review Protocol

Select Search Engines

Define Keyword

Collect Papers

Pick Relevant Papers

Keyword-based Search

A. Research Questions Based on our research goals, our systematic literature review mainly targets at answering the following three research questions: • RQ1: How to classify the current research on EA complexity cybernetics? • RQ2: What is state-of-the-art of EA complexity cybernetics? • RQ3: What can we learn from state-of-the-art to improve the future research? B. Research Tasks To answer the three research questions above, we need to analyze and decide research tasks. Figure 2 shows the three research questions and the corresponding research tasks. To answer RQ1, we need to analyze the contexts of target publications to find the common standards (such as background, method, etc.) , and then to categorize target publications into different categories by the common standards. To answer RQ2, we chart a landscape of EA complexity cybernetics by performing an extensive analysis, including

Fig. 3.

Search process

Define Review Protocol: The review protocol is designed to service for answering the research questions, it can provide guidelines for defining keywords and choosing search engines to ensure the quality of target publications. Select Search Engines: the search engines and databases we used include IEEE XPlore, ACM Digital Library, Springer, Elsevier and Google Scholar. Define Keyword: we define two basic group keywords, and select one from each basic group to combine the final keyword for searching papers. One basic group includes ”Enterprise Architecture” and ”EA”. The other group includes ”Complexity Cybernetics”, ”Complexity Management”, ”Complexity” and ”Complex”. Keyword-based Search: based on the final keywords above, we select search engines one by one to search papers. The time range of publications is set between 1999 to 2015.

Pick Relevant Papers: the search model may get some irrelevant papers. In this step, we should pick relevant papers by analyzing the title and keywords of papers. Collect Papers: in this step, we download the selected papers and collect some relevant data, such as type of publications. Finally, based on the search process, we get 33 relevant papers. IV. A NALYSIS AND R ESULTS In this section, we performed an analysis of the 33 papers at the aspects which the research tasks requested, in order to answer the first and second research questions. Subsection A shows the context analysis and its results, including the analysis of backgrounds and methods in the papers, mainly to answer the first research question. In Subsection B, an extensive analysis is conducted to answer the second research question. A. Context Analysis To answer the first research question (RQ1), we categorize EA complexity cybernetics, by backgrounds and methods which the papers relate to, based on the research tasks. By scanning the 33 papers and analyzing their backgrounds, we can categorize these 33 papers into 11 subgroups. The result is shown in Table 1. From the Table 1 we find the backgrounds of complexity cybernetics are various and scattered. To get a normative classification, as shown in Table 2, these 11 sub-groups can be categorized into 3 categories: (I) First category is a high-level or enterprise-level one with the core to be Enterprise Architecture and Information System, including 5 groups, i.e., Enterprise Architecture, Information System, Information Technology, Product Architecture and Service Oriented Architecture; (II) Second category is a midlevel or system-level one mainly based on Software Engineering and System Engineering, including 5 groups, i.e., Software Architecture, Software System, System of Systems, System Engineering and Complex System; (III) The final category is a low-level one based on Network of Enterprise, including Network of Enterprise. From the Table 2 we can find that although the research on the complexity cybernetics in enterprise-level accounts for 60.6% within 33 papers, still a large part of research focus on the complexity cybernetics in system-level. Another kind of categorization is based on the method/approach used to study complexity: either quantitative one or qualitative one. By analyzing the methods-models used in the 33 papers, we get a basic categorization, as shown in Table 3. From the Table 3 we can find that more than half of publications employ qualitative methods. The representative papers using qualitative method to control EA complexity include [11], [22] and [28]. Dag I.K. Sjoberg et al. [11] presented a method to assess the cost of IT complexity. The steps included Identify the Affected Systems, Extract the Affected Cost Base, Estimate the Change of Complexity Per System and Total economic impact. The core steps, i.e., Extract the Affected Cost Base and Estimate

TABLE I 11

SUB - GROUPS OF THE

33

Background

PAPERS .

Literature

Enterprise Architecture Information System Information Technology Product Architecture Service Oriented Architecture Software Architecture Software System System of System System Engineering Complex System Network of Enterprise

[1],[3],[4],[17],[19],[31],[33],[34],[35] [9],[36] [10],[13],[18],[24],[28] [12] [20],[23],[26] [25],[27],[30] [11],[22] [6] [5],[16] [14],[21] [15],[29],[32] TABLE II

T HE NORMATIVE CLASSIFICATION . Classification Enterprise Architecture/Information System Software Engineering/System Engineering Network of Enterprise total

Number

Percent

20 10 3 33

60.6% 30.3% 9.1% 100%

the Change of Complexity Per System, were mainly based on exports’ experiences. H. Kandjani et al. [22] used Extended Axiomatic Design (EAD) theory to study the complexity of Global Software Development (GSD) projects. By using EAD theory, they presented the Generalised Enterprise Reference Architecture (GERA) modeling framework to reduce the complexity of enterprise models. Besides, they [35] also studied the complexity of disaster management based on the method. K. Solic et al. [28] applied the Simple Iterative Partitions (SIP), presented by Sessions [2], to reduce the complexity of IT System. The SIP process included partitioning of the system, simplification of the system and iteration between subsets of the system. Besides, representative papers, such as [1], [23] and [27], use quantitative method to manage EA complexity. A. Schuetz et al. [1] proposed a metric to quantify complexity in EA. They considered complexity as tuple of number and heterogeneity of the relations of an EA [1]. L. Liu et al. [23] presented a simulation-based approach to calculate the complexity in Service-oriented Computing. S. Sengupta et al. [27] presented the complexity metrics based on Component Architecture Graph (CAG) to calculate component architecture complexity. The metrics belonged to to different levels, i.e., component level metrics and architecture Level metrics. Due to the limited length of the paper, the formulas of these metrics are not listed. B. Extensive Analysis To answer the second research question (RQ2), we preformed an extensive analysis by analyzing the data discussed in research tasks, including distribution of papers over time, their regional distribution, type of publication, and authorgroup distribution. The number of EA complexity cybernetics publications

TABLE III T HE CLASSIFICATION OF 33 PAPERS BY DIFFERENT METHODS . Method

TABLE IV T YPE OF P UBLICATION D ISTRIBUTION .

Number

Percent

Representative

Type of publication

20 13

60.6% 39.4%

[11],[22],[28] [1],[23],[27]

Conferences Journals Books Reports total

Qualitative Quantitative

over year is shown in Figure 4. From the Figure 4 we can find that (1) 33 papers are all published during the period from 2002 to 2015; and (II) the publications from 2009 to 2014 are more than other period.

Griffith University TU Munchen BPPIMT

6

Percent

22 7 2 2 33

66.6% 21.2% 6.1% 6.1% 100%

TABLE V T HE T OP -3 P RODUCTIVE AUTHORS G ROUPS D ISTRIBUTION . Authors groups

7

Number

Publications

Number

[15],[21],[22],[32],[34] [1],[4],[5],[12] [20],[27]

5 4 2

5 4 3 2 1

Fig. 4.

2015

2014

2013

2012

2011

2010

2009

2008

2007

2005

2004

2002

0

Distribution of Papers: Over Time

In Figure 5, we show the regional distribution of the 33 papers. The 33 papers come from 11 countries, and the top-3 most productive countries are USA, Germany and Australia. More than half of all publications come from these three countries. Argentina, 1 USA, 7

Australia,, 6

UK, 1 Canada, 3

Sweden 1 Sweden, Norway, 1

China, 2

Netherlands ,2

Croatia,, 1 India, 2 Germany, 6

Fig. 5.

Regional Distribution

Table 4 shows the type of publication distribution. We can find a vast amount of papers are published in conference proceedings, and 7 papers are published in journals. Besides, 2 results are from books and another two are working reports. The top-4 most productive types of publication include: Hawaii International Conference on System Sciences (HICSS), ACM SIGSOFT Software Engineering (SIGSOFT),

Computer-Human Interaction for Management of Information Technology (CHIMIT) and Americas Conference on Information Systems (ACIS). HICSS contributed 3 papers and the other three types of publication contributed 2 papers to the total 33 papers, while others only contributed one. Besides, we analyzed the author-group distribution. Majority authors groups are universities, while only three papers’ authors come from industrial companies. As showed in Table 5, the top-3 productive organizations are Griffith University (in Australia), Technische Universitaet Munchen (in Germany) and B.P. Poddar Institute of Management and Technology (BPPIMT) (in India). To summarize, research of EA complexity cybernetics has been quite hot in recent years, and many academic and industrial organizations or research groups have paid lots of attention to this domain, especially some research organizations from Europe, Oceania and USA. V. D ISCUSSIONS In this section, we mainly answer the third research question (RQ3). With a detailed analysis of the research on EA complexity cybernetics above, there are some conclusions. • The research taxonomy of EA complexity cybernetics is not unified. As the domain of EA complexity cybernetics is still young, the research taxonomy is not unified, and a unified framework is needed to advance the future research. Although our work make a basic taxonomy of the current research by different backgrounds and methods the papers referred to, more efforts still need to be added to the research taxonomy. • The research groups is limited, while co-work between different research groups may be helpful. From the current state of EA complexity cybernetics, some research groups in Europe, Oceania and USA keep relative advantage compared to others in the field. Besides, high quality publications are yet few. Co-work between different research groups may be helpful.

Conceptual models and research based on conceptual models shape a new important trend. Although there are many researches on EA complexity, conceptual model of EA complexity is not yet well-unified. The definition and scope of EA complexity need to be ascertained and unified. Besides, research method and process based on conceptual model need to be focused in future research. •

VI. C ONCLUSIONS This paper presents a systematic literature analysis of EA complexity cybernetics. We conduct an analysis of 33 papers to discuss the research taxonomy, state-of-the-art, and future directions of EA complexity cybernetics. This research illustrates a basic landscape of EA complexity cybernetics. This sort of work will help us to better understand the current state-of-the-art of EA complexity cybernetics. Future related work could be conducted with a focus on conceptual models and research method (based on conceptual model) of EA complexity cybernetics. R EFERENCES [1] A. Schuetz, T. Widjaja and J. Kaiser, “Complexity in Enterprise Architectures - Conceptualization and Introduction of a Measure from a System Theoretic Perspective,” Proceedings of the 21st European Conference on Information Systems, Utrecht, The Netherlands, 2013. [2] R. Sessions, Simple Architecture for Complex Enterprises, Microsoft Press, 2008. [3] A.W. Schneider, M. Zec and F. Matthes, “Adopting Notions of Complexity for Enterprise Architecture Management,” Twentieth Americas Conference on Information Systems. Savannah, 2014. [4] A.W. Schneider, T. Reschenhofer, A. Schutz and F. Matthes,“Empirical Results for Application Landscape Complexity,” System Sciences (HICSS), 48th Hawaii International Conference on, pp. 4079-4088, 2015. [5] M. Maurer, R. Schneller and M. Omer, A survey on complexity management in systems engineering, 2014. [6] P.N. Lowe and M.W. Chen, System of Systems Complexity Modeling and Simulation Issues, Working Report, The Boeing Company, 2008. [7] B. Kitchenham, Procedures for Performing Systematic Reviews. keele 33, United Kingdom, 2004. [8] C. Okoli and K. Schabram, “A Guide to Conducting a Systematic Literature Review of Information Systems Research,” Social Science Electronic Publishing. 10(26), 2010. [9] B.L. Marcolin and A. Ross, “Complexities in IS sourcing: equifinality and relationship management,” Data Base for Advances in Information Systems, vol. 36, no. 4, pp. 29-46, 2005. [10] B. Lin, A.B. Brown and J.L. Hellerstein, “Towards an understanding of decision complexity in IT configuration,” Autonomic Computing, ICAC ’06. IEEE International Conference on, pp. 279-282, 2006. [11] Dag I.K. Sjoberg, E. Odberg and B. Warlo, “The challenge of assessing and controlling complexity in a large portfolio of software systems,” PROFES, June 21-23, Limerick, Ireland, 2010. [12] C. Daniilidis, D. Hellenbrand, W. Bauer and U. Lindemann, “Using structural complexity management for design process driven modularization,” IEEE International Conference on Industrial Engineering and Engineering Management, pp. 595-599, 2011. [13] G. Robiolo, “How Simple is It to Measure Software Size and Complexity for an IT Practitioner,” Empirical Software Engineering and Measurement , 2011 International Symposium on, pp. 40-48, 2011. [14] G.D. Snook, “A general theory of complex living systems: Exploring the demand side of dynamics,” Complexity, vol. 13, no. 6, pp. 12-20, 2007. [15] H. Kandjani, “Enterprise Architecture Cybernetics for Collaborative Networks: Reducing the Structural Complexity and Transaction Cost Via Virtual Brokerage,” Concurrent Engineering, pp. 1233-1239, 2012.

[16] A.G. Hessami and N. Karcanias, “Integration of operations in process systems: Complexity and emergent properties,” INCOSE International Symposium, pp. 466-471, 2011. [17] E.J. Roest, “The Relationship Between enterprise architecture, business complexity and business performance,” Master Thesis, Business Information Technology, School of Management and Governance, University of Twente. 2014. [18] J.L. Lentz and T.M. Bleizeffer, “IT ecosystems: evolved complexity and unintelligent design,” CHIMIT’07, March 30-31, Cambridge, MA, U.S.A., 2007. [19] J. Saat, S. Aier and B. Gleichauf , “Assessing the Complexity of Dynamics in Enterprise Architecture Planning-Lessons from Chaos Theory,” Proceedings of the Fifteenth Americas Conference on Information Systems, San Francisco, California, 2009. [20] J. Chanda, S. Sengupta, A. Kanjilal and S. Bhattacharya, “SCAG: a graphical approach to measure the complexity of the SOA application,” Acm Sigsoft Software Engineering Notes, vol. 36, no. 5, pp. 1-6, 2011. [21] H. Kandjani, P. Bernus and S. Nielsen, “Enterprise Architecture Cybernetics and the Edge of Chaos: Sustaining Enterprises as Complex Systems in Complex Business Environments,” Proceedings of the 46th Hawaii International Conference on System Sciences, Wailea, HI, USA, pp.3858-3867, 2013. [22] H. Kandjani, P. Bernus and L. Wen, “Enterprise Architecture Cybernetics for Complex Global Software Development: Reducing the Complexity of Global Software Development Using Extended Axiomatic Design Theory,” IEEE Seventh International Conference on Global Software Engineering, pp.169-173, 2012. [23] L. Liu, S. Thanheiser and H. Schmeck, “Assessing Complexity Of Service-Oriented Computing Using Learning Classifier Systems,” SAC ’09 Proceedings of the 2009 ACM symposium on Applied Computing, pp. 2170-2171, 2009. [24] M. Mocker, “What Is Complex About 273 Applications - Untangling Application Architecture Complexity in a Case of European Investment Banking,” 47th Hawaii International Conference on System Sciences, pp. 1-14, 2009. [25] M. Alsharif, W.P. Bond and T. Al-Otaiby, “Assessing the complexity of software architecture,” Acm Southeast Regional Conference, pp. 98103, 2004. [26] Q. Zhang and X. Li, “Complexity Metrics for Service-Oriented Systems,” Second International Symposium on Knowledge Acquisition and Modeling, pp. 375-378, 2010. [27] S. Sengupta, A. Kanjilal, and S. Bhattacharya, “Measuring complexity of component based architecture: a graph based approach,” Acm Sigsoft Software Engineering Notes, vol. 36, no. 1, pp. 1-10, 2011. [28] K. Solic, D. Sebo and F. Jovic, “Methodology for complexity reduction of IT system (adjustment of the sessions’ methodology),” Mipro, pp. 1528 - 1531, 2011. [29] X. Sun, S.G. Rao and G.G. Xie, “Modeling Complexity of Enterprise Routing Design,” CoNEXT’12, pp. 85-96, 2012. [30] Y. Liu and I. Traore, “Complexity Measures for Secure ServiceOriented Software Architectures,” Predictor Models in Software Engineering, International Workshop on, pp. 11-11, 2007. [31] H. Lee, J. Ramanathan, Z. Hossain, P. Kumar, B. Weirwille and R. Ramnath, “Enterprise Architecture Content Model Applied to Complexity Management While Delivering IT Services,” 2014 IEEE International Conference on Services Computing (SCC), pp. 408-415, 2014. [32] H. Kandjani, L. Wen and P. Bernus, “Enterprise Architecture Cybernetics for Global Mining Projects- Reducing the Structural Complexity of Global Mining Supply Networks via Virtual Brokerage,” Advanced Materials Research, pp. 634-638:3339-3345, 2013. [33] S.L. James and Donald W. de Guerre, “Enterprise-in-Environment Adaptation Enterprise Architecture and Complexity Management,” A Systemic Perspective to Managing Complexity with Enterprise Architecture, 2014. [34] H. Kandjani and P. Bernus, “Enterprise Architecture Cybernetics for Complex Disaster Management,” 15th International Conference of Enterprise Information Systems, 2013. [35] H. Li and T.J. Williams, “Management of complexity in enterprise integration projects by the PERA methodology,” Journal of Intelligent Manufacturing, vol. 13, no. 6, pp. 417-427, 2002. [36] Y. Yuan, P. Wang, X.Y. Song and W.D. Zhang, “A Complexity Analysis on Enterprise Information Ecosystem,” Journal of Jilin Institute of Architecture Civil Engineering, vol. 26, no. 4, pp. 68-71, 2009.