Performance Evaluation of Hierarchical Publish-Subscribe Monitoring Architecture for Service-Oriented Applications

TL;DR

This paper evaluates a hierarchical publish-subscribe monitoring architecture for service-oriented applications, focusing on performance aspects like load, latency, and scalability using XML-based reports.

Contribution

It provides a performance analysis of a tree-topology publish-subscribe monitoring system with XML reports, addressing scalability and efficiency issues.

Findings

XML formatting impacts processing and network load

Performance depends on number of nodes and system load

Latency increases with system size and load

Abstract

Contemporary high-performance service-oriented applications demand a performance efficient run-time monitoring. In this paper, we analyze a hierarchical publish-subscribe architecture for monitoring service-oriented applications. The analyzed architecture is based on a tree topology and publish-subscribe communication model for aggregation of distributed monitoring data. In order to satisfy interoperability and platform independence of service-orientation, monitoring reports are represented as XML documents. Since XML formatting introduces a significant processing and network load, we analyze the performance of monitoring architecture with respect to the number of monitored nodes, the load of system machines, and the overall latency of the monitoring system.