Masterarbeit von Ludwig Achhammer

Assessing the Cost and Benefit of a Microservice Landscape Discovery Method

Motivation & problem statement:
As today’s enterprises increasingly rely on their information systems and IT infrastructure, Enterprise Architecture Management (EAM) has become a strategic discipline for improving Business-IT alignment and support business goals [11]. An Enterprise Architecture (EA) models business and IT artifacts and puts them into relation with one another. Having a thorough understanding of the inter-dependencies between these elements, strategic decisions can be derived more easily to transform the enterprise architecture into the desired direction in line with business needs [8]. This process of transformation is usually accompanied by modeling a current state, one or more planned states and a long term target state of the enterprise architecture [38].

A critical success factor to reliably capture the current state of an enterprise architecture is to build upon data that is in sync and accurately reflects the reality. The documentation of the as-is enterprise architecture, therefore, constitutes a major challenge for the EAM discipline as being found by practitioners and researchers [15].

In the context of the growing adoption of agile software development methodologies and the increased usage of microservice-based architectures, today’s IT landscapes are exposed to rapid architectural changes. As a consequence, EA documentation processes are challenged all the more to keep pace with continuously changing and growing complexity of the IT landscape. Current research and practitioners are in search of an appropriate solution to this challenge and to automate the documentation and maintenance of EA models. Several solution approaches have been proposed (e.g. based on Network Scanners [16], cloud platforms [7], Enterprise Service Buses [4], cloud platforms [8] and machine learning [23]) by literature. However, most of this solutions do not cover the EA business layer and have not been reflected within real-case enterprise environments.

As part of the chair’s research in this area, a novel approach to automate the EA model documentation and maintenance process within microservice-based environments was developed. The two key enabling technologies of this approach are (1) the use of cloud platform and distributed tracing data which allows the discovery of microservices and their inter-dependencies at runtime and (2) the integration into CI/CD pipelines. This thesis puts the focus on automated, pipeline-driven EA documentation. The approach uses software deployment processes as a trigger for the automated EA documentation process. Application related data collected at runtime is completed to a multi-layer EA model by the help of static information coupled to the software artifact using a configuration file.

This work follows up on a proof of concept of this novel approach and further develops it into an integral part of the EA processes within a large German enterprise. This includes the roll-out to a real-case environment and the integration into existing processes. Based on this, the goal is to critically assess the cost and benefit of this solution approach from several perspectives. The assessment should deliver prove about the feasibility and the capabilities of the suggested solution in practical use.

Research questions:
As part of this work the following research questions will be handled and answered:
RQ1: How can the suggested solution be integrated into agile development and what challenges do occur?
RQ2: What EA model elements should be documented and to what degree can this be automated using the solution approach?
RQ3: What are the solution’s integration costs and value propositions for Enterprise Architecture Management?

RQ1 targets at the integration and of the suggested solution approach into a real-case ecosystem (processes, tools and actors) and evaluate its behavior in productive use. The goal is to identify practical challenges that arise and provide recommendations on how they might be overcome.

RQ2 will be answered as part of the case study conducted at the industry partner. By assessing the solution’s capabilities within a real-case enterprise context, it will be clarified to what degree the target EA meta model can be discovered automatically. This will also reveal the limitations of the suggested approach.

RQ3 will be answered by evaluating the suggested solution approach from multiple perspectives and evaluation criteria. Judgments are based on findings made throughout the case study, an analysis of cost and savings as well as feedback obtained during a series of expert interviews. The goal is to draw a cost/benefit ratio about the suggested solution approach.


Literature:
[11] M. Farwick, C. M. Schweda, R. Breu, and I. Hanschke, “A situational method for semi-automated enterprise architecture documentation,” Software & Systems Modeling, vol. 15, no. 2, pp. 397–426, 2016, issn: 1619-1366. doi: 10.1007/s10270-014-0407-3.

[8] M. Farwick, B. Agreiter, R. Breu, S. Ryll, and I. Hanschke, “Requirements for automated enterprise architecture model maintenance - a requirements analysis based on a literature review and an exploratory survey,” 2011, pp. 325–337.

[38] The TOGAF standard, Version 9.2. Zaltbommel: Van Haren Publishing, 2018, isbn: 9789401802833.

[16] H. Holm, M. Buschle, R. Lagerström, and M. Ekstedt, “Automatic data collection for enterprise architecture models,” Software & Systems Modeling, vol. 13, no. 2, pp. 825–841, 2014. doi: 10.1007/s10270-012-0252-1.

[7] M. Farwick, B. Agreiter, R. Breu, M. Häring, K. Voges, and I. Hanschke, “Towards living landscape models: Automated integration of infrastructure cloud in enterprise architecture management,” in 2010 IEEE 3rd International Conference on Cloud Computing: A Requirements Analysis based on a Literature Review and an Exploratory Survey, IEEE, 2010, pp. 35–42, isbn: 978-1-4244-8207-8. doi: 10.1109/CLOUD.2010.20.

[15] M. Hauder, F. Matthes, and S. Roth, “Challenges for automated enterprise architecture documentation,” in Trends in enterprise architecture research and practicedriven research on enterprise transformation : 7th Workshop, TEAR 2012, and 5th Working Conference, PRET 2012 ; held at the Open Group Conference 2012, Barcelona, Spain, October 23-24, 2012 ; proceedings, vol. 131, Springer, 2012, pp. 21–39, isbn: 978-3-642-34162-5. doi: 10.1007/978-3-642-34163-2_2.

[4] M. Buschle, M. Ekstedt, S. Grunow, M. Hauder, F. Matthes, and S. Roth, “Automating enterprise architecture documentation using an enterprise service bus,” in AMCIS, 2012.

[23] J. Landthaler, Ö. Uluda˘ g, G. Bondel, A. Elnaggar, S. Nair, and F. Matthes, “A machine learning based approach to application landscape documentation,” in The Practice of Enterprise Modeling, ser. Lecture Notes in Business Information Processing, R. A. Buchmann, D. Karagiannis, and M. Kirikova, Eds., vol. 335, Cham: Springer International Publishing, 2018, pp. 71–85, isbn: 978-3-030-02301-0. doi: 10.1007/978-3-030-02302-7_5.