Self-Adaptive Microservice-based Systems -- Landscape and Research Opportunities

TL;DR

This paper surveys the current state of self-adaptation techniques in microservice systems, analyzing 21 studies to identify trends, gaps, and future research directions in enhancing resilience, performance, and evolution.

Contribution

It provides a systematic mapping of existing research on self-adaptation in microservices, highlighting predominant strategies, focus areas, and identifying gaps for future exploration.

Findings

Most studies focus on the Monitor phase of adaptation.

Self-healing is the most addressed property.

Reactive adaptation strategies are predominant.

Abstract

Microservices have become popular in the past few years, attracting the interest of both academia and industry. Despite of its benefits, this new architectural style still poses important challenges, such as resilience, performance and evolution. Self-adaptation techniques have been applied recently as an alternative to solve or mitigate those problems. However, due to the range of quality attributes that affect microservice architectures, many different self-adaptation strategies can be used. Thus, to understand the state-of-the-art of the use of self-adaptation techniques and mechanisms in microservice-based systems, this work conducted a systematic mapping, in which 21 primary studies were analyzed considering qualitative and quantitative research questions. The results show that most studies focus on the Monitor phase (28.57%) of the adaptation control loop, address the self-healing property (23.81%), apply a reactive adaptation strategy (80.95%) in the system infrastructure level (47.62%) and use a centralized approach (38.10%). From those, it was possible to propose some research directions to fill existing gaps.