Patterns for Monitoring Parallel Processes

Monitoring provides well established means to track the exchanged messages in cross-organizational processes. Complex Event Processing (CEP) is a technology used by monitoring systems that aims to process incoming events and generate high-level information. Current implementation of CEP has performance issues when the processing of events involves calling a composition of services to generate the required information. In this paper we develop a conceptual framework that relies on a set of patterns that maximize the parallel processing of incoming events. These patterns are used to split the composition of services into a set of blocks that can be executed in parallel while ensuring an effective correlation of the intermediate results. Our approach significantly decreases the processing time without requiring too many resources. We implemented a prototype of our framework and conducted performance evaluation in comparison with existing software.