Contract-Drift Detection in Event-Driven Microservices Using Runtime Message Semantics: A Literature Synthesis

Abstract

Event-driven microservice architectures largely dispense with synchronous interfaces, in which producer and consumer services are bound to a shared interface at compile time, and rely instead on asynchronous message contracts that may evolve between services without explicit notice — a condition termed contract drift. This paper synthesises fourteen peer-reviewed and archival papers published between 2017 and 2022 across five domains: microservice architecture, consumer-driven contract testing, semantic versioning, runtime-verification theory, and distributed observability. The synthesis finds that prevailing assurance mechanisms operate predominantly offline and pre-deployment, and therefore cannot detect drift that emerges once a service is in production. Building on an established runtime-verification taxonomy, the paper proposes a structural-similarity measure and a type-agreement measure for quantifying the extent of contract drift; both are computed in flight against a registered baseline contract, and an alert is raised when their weighted combination exceeds a configurable threshold. The available evidence indicates pronounced recent interest in the problem — six of the fourteen papers synthesised were published in 2022 — and suggests that a decentralised, broker-side monitoring architecture is a viable design approach. Open challenges remain in baseline governance, behavioural drift detection, and the absence of standardised benchmarks for comparing runtime contract-monitoring techniques.