Formal Verification Frameworks for Cloud-Native Microservices

Abstract

Microservices that are cloud-native allow distributed systems to scale, but present problems in assuring correctness, reliability, and fault resilience because of dynamic scaling and the complexity of service interactions. The presented paper suggests a formal verification framework that consists of a combination of specifications of temporal logic, model checking and modeling of service interactions graphs to study the behavior of the system and confirm the correctness properties. Service states, dependencies and scaling dynamics are modeled in a mathematical system. The framework identifies deadlocks, provides liveness and safety, and a framework of fault propagation between services. Experimental analysis shows that the accuracy of deadlock detection is increased to 96.7% and fault detection to 95.4% with 94.6% verification coverage and time of 96 ms to detect faults is minimized as compared to the conventional testing and monitoring methods in dynamic microservice settings.