Performance evaluation of a distributed application performance monitor

Abstract

The Distributed Measurement System (DMS) is a software-based measurement infrastructure for monitoring the performance of distributed application systems. In this paper we evaluate DMS in two configurations: a monitor for quality of service and a collector for model building parameters. Three distributed application workload types are defined and a model for DMS is given. The model parameters for DMS are based on measurement data from an implementation of DMS for the Open Software Foundation's Distributed Computing Environment. We use the model with our workloads to consider the impact of DMS on processor and network utilization and on workload responsiveness. We show how the various factors that control DMS affect its overhead. Lastly, the scalability of DMS is considered for large distributed environments. Our results indicate that DMS is well suited for monitoring QoS and supporting workload characterization for model building.