SelfTalk for Dena: Query Language and Runtime Support for Evaluating System Behavior

摘要

We introduce SelfTalk, a novel declarative language that allows users to query and understand the status of a large scale system. SelfTalk is sufficiently expressive to encode an administrator's high level hypotheses/expectations about normal system behavior, such as, "I expect that the throughputs across all system components are linearly correlated". SelfTalk works in conjunction with Dena, a runtime support system designed to help system administrators detect the root cause of system misbehavior quickly and accurately. Given a user hypothesis, Dena instantiates and validates it using actual monitored data within specific system contexts. We evaluate Dena by posing several hypotheses about system behavior and querying Dena to diagnose anomalies in a virtual storage system. We find that Dena can automatically validate the system performance based on the user hypotheses and also accurately diagnose system misbehavior.