First-order monitors analyze data-carrying event streams. When event streams are generated by distributed systems, it may be difficult to ensure that events arrive at the monitor in the right order. We develop a new monitoring tool for metric first-order temporal logic, called TimelyMon, that can process out-of-order events. Using the stream processing framework Timely Dataflow, TimelyMon also supports parallelized monitoring. We demonstrate TimelyMon's good performance and scalability on synthetic and real-world benchmarks.
Paper draft