A Multi-variant Benchmark for Microservice Systems in Software Engineering Research

Microservice architecture became the mainstream for cloud-native systems. While many microservice system benchmarks have been introduced to the scientific community, there is still a notable gap since the benchmarks do not offer architectural variants of the same system with identical functionality. For instance, research engaged mono-to-micro decomposition, but proposed methods use textbook or tutorial systems. Moreover, different microservice granularities in the system provide different architectural trade-offs. This paper extends an established microservice benchmark with two new variants, including a system monolith and a version with 20 microservices. Equivalent functionality is validated across the three benchmark variants.