Exploring Maintainability Assurance Research for Service- and Microservice-Based Systems: Directions and Differences

Justus Bogner, Adrian Weller, Stefan Wagner, Alfred ZimmermannMicroservices 2019 · DOI

TL;DR

This paper presents a systematic literature review (SLR) of research on maintainability assurance for service- and microservice-based systems. It categorizes existing approaches and reveals a significant imbalance, with the vast majority of research focusing on older Service-Oriented Architectures (SOA) and very little dedicated to the newer microservices paradigm. The findings highlight opportunities to adapt SOA research for microservices and suggest future research directions.

What Problem Does This Paper Address?

As software systems become increasingly central to businesses, ensuring their long-term maintainability and evolvability is crucial. Service-oriented architectures (SOA) and microservices have become popular for building scalable and adaptable systems. However, the landscape of research on how to assure the maintainability of these systems is fragmented and difficult to navigate. Specifically, while SOA has been extensively studied, the newer microservices architecture, despite its growing adoption, has a much smaller body of dedicated research concerning maintainability assurance. This paper aims to bring clarity by systematically reviewing and categorizing existing research, identifying trends, and highlighting gaps, particularly for microservices.

Methods

The authors conducted a systematic literature review (SLR) following established guidelines. Their process involved:

  1. Defining Research Questions: Four key questions guided the review, focusing on categorization, distribution of research, research directions, and differences between SOA and microservices.
  2. Designing a Detailed Protocol: This included defining data sources (IEEE Xplore, ACM Digital Library, Springer Link, ScienceDirect, and Google Scholar for snowballing), search terms, and inclusion/exclusion criteria.
  3. Retrieving and Filtering Studies: An initial search yielded 1000 papers, which were filtered down to 122 based on relevance.
  4. Snowballing: Forward citation searching on the filtered set identified an additional 806 publications.
  5. Final Selection and Analysis: After applying filters and merging results, a final set of 223 primary studies was selected. These studies were then analyzed and categorized using a three-dimensional taxonomy.
  6. Categorization Scheme:
    • Architectural: SOA, Microservices, or Both.
  7. Methodical: Case/Field/Empirical Study, Literature Study, Model/Taxonomy, Process/Method, Reference Architecture/Tool.
  8. Thematic: Specific subfields of maintainability assurance (e.g., Evolution Management, Maintainability Metrics & Prediction, Change Impact & Scenarios, etc.).

The categorization and analysis were performed iteratively by two researchers to ensure rigor.

Datasets

Not discussed in the paper. The paper focuses on a systematic review of existing literature, not on the creation or evaluation of new datasets.

Key Insights and Ideas

  • Microservices Research Lag: The most striking finding is the significant underrepresentation of microservices in maintainability assurance research. While SOA research is abundant, dedicated work on microservices is scarce, suggesting a critical gap.
  • Adaptability of SOA Research: Despite the architectural differences, many SOA-centric maintainability assurance techniques are likely adaptable to microservices, especially those related to RESTful services.
  • Dominant Thematic Areas: For both SOA and microservices, "Evolution Management" and "Maintainability Metrics & Prediction" are the most popular thematic areas, indicating a strong interest in managing system evolution and quantifying maintainability.
  • Methodological Preferences: Case, field, or empirical studies are prevalent, suggesting a practical focus on demonstrating and evaluating approaches in real-world or simulated scenarios.
  • Architecture Recovery for Microservices: A notable exception to the general lack of microservices research is in "Architecture Recovery and Documentation," where microservices are relatively well-represented, highlighting the challenges of understanding decentralized architectures.
  • Antipatterns: Definition vs. Detection: For antipatterns, SOA research has moved towards automatic detection, while microservices research is still largely focused on defining and conceptualizing them.

Key Results

  • 223 Primary Studies: The SLR identified 223 relevant publications from 2007 to 2018.
  • Architectural Distribution:
    • 90% (199 studies) focused exclusively on SOA.
  • Only ~11% (24 studies) focused on microservices (either exclusively or in combination with SOA).
  • Even in a more recent subset (2014-2018), microservices-related publications accounted for only ~25%.
  • Methodical Categories:
    • Case, Field, or Empirical Study: 63%
    • Process or Method: 49%
    • Model or Taxonomy: 34%
    • Reference Architecture or Tool: 32%
    • Literature Study: 11%
  • Thematic Categories:
    • Evolution Management: 25%
    • Maintainability Metrics & Prediction: 25%
    • Change Impact & Scenarios: 17%
    • Antipatterns & Bad Smells: 9%
    • Service Identification & Decomposition: 9%
    • Patterns: 6%
    • Model-Driven Approaches: 5%
    • Architecture Recovery & Documentation: 4%
  • Microservices Specifics: In thematic categories, microservices research was most prominent in Architecture Recovery & Documentation (5 of 9 papers). For Antipatterns & Bad Smells, microservices research focused on definition, while SOA research focused on detection.

Limitations and Future Work

Limitations:

  • Search Strategy: The authors acknowledge that the search strategy might have missed relevant papers due to keyword limitations or database sorting.
  • Result Set Size: Limiting the initial search to the first 250 results per publisher (1000 total) might have excluded relevant studies, especially if search engines' relevance sorting changes.
  • Exemplary Studies: The selection of exemplary studies to illustrate research directions was based on the authors' perception of relevance.

Future Work:

  • Deeper Dives: Conduct literature studies for individual thematic categories to gain more in-depth insights.
  • Orchestration vs. Choreography: Analyze differences in maintainability assurance research between SOA (often orchestration-heavy) and microservices (often choreography-based).
  • Replication for Microservices: Replicate this SLR specifically for microservices in a few years, given the rapid growth in this area.
  • Empirical Studies: Investigate potential negative maintainability impacts of microservices (e.g., knowledge exchange, team synchronization, heterogeneity) through empirical studies.

Summary updated

BibTeX
@inproceedings{bogner2019exploring,
  author = {Justus Bogner and Adrian Weller and Stefan Wagner and Alfred Zimmermann},
  title = {{Exploring Maintainability Assurance Research for Service- and Microservice-Based Systems: Directions and Differences}},
  booktitle = {Microservices},
  year = {2019},
  doi = {10.4230/oasics.microservices.2017-2019.3},
}
Figure