Authors:
Josef Schiefer
1
and
Carolyn McGregor
2
Affiliations:
1
IBM Watson Research Center, United States
;
2
Centre for Advanced Systems Engineering, University of Western Sydney, Australia
Keyword(s):
Event Stream Processing, Event Correlation, Business Process Monitoring
Related
Ontology
Subjects/Areas/Topics:
Coupling and Integrating Heterogeneous Data Sources
;
Data Warehouses and OLAP
;
Databases and Information Systems Integration
;
Deductive, Active, Temporal and Real-Time Databases
;
Enterprise Information Systems
;
Enterprise Resource Planning
;
Enterprise Software Technologies
;
Simulation and Modeling
;
Simulation Tools and Platforms
;
Software Engineering
Abstract:
With the increasing demand for real-time information on critical performance indicators of business processes, the capturing, transformation and correlation of real-world events with minimal latency are a prerequisite for improving the speed and effectiveness of an organization's business operations. Events often include key business information about their relationship to other events that can be utilized to collect relevant event data for the calculation of business performance indicators. In this paper we introduce an approach for correlating events of business processes that uses correlation sessions to represent correlation knowledge. Correlation sessions facilitate the processing of data across multiple events and thereby enable a calculating of business metrics in near real-time. The benefit over existing approaches is that it is tailored to instrument business processes and business applications that may operate in a heterogeneous software environment. We propose a Java-based
, container-managed environment which provides a distributed, scalable, near-real time processing of events and which includes a correlation service that effectively manages correlation sessions. We also show a complete example that illustrates how correlation sessions can be utilized for computing the cycle time of business processes.
(More)