An Approach for Evaluating the Potential Impact of Anti-Patterns on Microservices Performance
The microservices architectural style has been increasingly adopted in recent years due to its advantageous characteristics. While there exist design patterns for microservices that are theoretically beneficial for ensuring performance (and anti-patterns that should be avoided), they do not always lead to the expected impact. Some designs are solutions for achieving quality properties other than performance, and while they might be categorized as anti-patterns in theory, their influence on system performance can be minimal, depending on the context. Architects find it hard to reason about these trade-offs and the impact that design decisions will have on performance before implementing them. To solve this problem, in this paper, we propose an approach that enables rapid evaluation of high-level architectural models by combining both static and dynamic analysis. The static analysis identifies the design anti-patterns that are known to hinder system performance and guides the architect in reasoning about these design decisions. The approach then generates source code for the system under study based on the architectural design model for a subsequent dynamic analysis to assess whether the statically detected anti-patterns do indeed have a negative effect on the performance of the analyzed system.