Performance Modeling and Benchmarking of Event-Based Systems
1st Edition
Published on 30. August 2011
Book
Paperback/Softback
184 pages
978-3-86844-353-0 (ISBN)
Description
Event-based systems (EBS) are increasingly used as underlying technology in many mission critical areas and large-scale environments, such as location-based services and environmental monitoring. They are typically highly distributed and data intensive with stringent requirements for performance and scalability.
To address these requirements, this book focuses on approaches to benchmark EBS as well as performance modeling methodologies. It introduces the reader to the field of EBS and discusses differences compared to traditional software in fundamental aspects such as underlying communications paradigm and performance.
We give a comprehensive overview of the first industry standard benchmark for message-oriented middleware (MOM), SPECjms2007, and of jms2009-PS, a research benchmark for publish/subscribe-based communication based on the SPECjms2007 workload. We present a methodology to analyze the performance of middleware using benchmarks and illustrate our approach in a case study on a commercial MOM.
To build performance models targeting EBS, we introduce a comprehensive modeling methodology. Several performance modeling patterns and extensions for Queueing Petri nets are proposed. The accuracy of our methodology is illustrated in two case studies applied to a distributed EBS as well as a MOM.
To address these requirements, this book focuses on approaches to benchmark EBS as well as performance modeling methodologies. It introduces the reader to the field of EBS and discusses differences compared to traditional software in fundamental aspects such as underlying communications paradigm and performance.
We give a comprehensive overview of the first industry standard benchmark for message-oriented middleware (MOM), SPECjms2007, and of jms2009-PS, a research benchmark for publish/subscribe-based communication based on the SPECjms2007 workload. We present a methodology to analyze the performance of middleware using benchmarks and illustrate our approach in a case study on a commercial MOM.
To build performance models targeting EBS, we introduce a comprehensive modeling methodology. Several performance modeling patterns and extensions for Queueing Petri nets are proposed. The accuracy of our methodology is illustrated in two case studies applied to a distributed EBS as well as a MOM.