Search papers, labs, and topics across Lattice.
This paper introduces PERF, a formal framework for predicting the performance of distributed systems under various fault scenarios. The framework uses a fault injector library and model compositions to integrate the system and fault injector into a unified model. The unified model is then analyzed using statistical methods to predict performance properties like throughput and latency, demonstrating accurate predictions consistent with real deployments.
Accurately predicting distributed system performance under diverse fault conditions is now possible directly from formal designs, thanks to a new automated framework.
Today's distributed systems operate in complex environments that inevitably involve faults and even adversarial behaviors. Predicting their performance under such environments directly from formal designs remains a longstanding challenge. We present the first formal framework that systematically enables performance prediction of distributed systems across diverse faulty scenarios. Our framework features a fault injector together with a wide range of faults, reusable as a library, and model compositions that integrate the system and the fault injector into a unified model suitable for statistical analysis of performance properties such as throughput and latency. We formalize the framework in Maude and implement it as an automated tool, PERF. Applied to representative distributed systems, PERF accurately predicts system performance under varying fault settings, with estimations from formal designs consistent with evaluations on real deployments.
Tsinghua AI