G. Din, "A TTCN3 based Benchmark for BPEL Engines," ETSI TTCN-3 User Conference 2008: Madrid (Spain), June 3-6, 2008.
Added by: Deleted user (22 Aug 2008 10:56:13 Europe/Berlin) Last edited by: Deleted user (17 Sep 2008 15:50:01 Europe/Berlin)
|Resource type: Conference Paper
BibTeX citation key: Dine
View all bibliographic details
Publisher: MTP, ETSI (Madrid (Spain))
Collection: ETSI TTCN-3 User Conference 2008
Views index: %
Popularity index: 1.75%
Benchmarking is a complex conceptual and technical activity to answer questions related to the performance of a system. Some of the typical questions a benchmark test answers are: does the application respond quickly enough for the intended users, will the application handle the expected user load and beyond, how many users before it gets “slow” or stops working? Benchmarking requires not only concepts, methods and tools but also a broad understanding of software and systems.
This presentation describes the design of a benchmark tool capable of evaluating the performance of Business Process Execution Language (BPEL) Engines for Web Services for configurable workloads. The approach describes the concepts of call flows, test procedures, metrics and design objectives as a general method to describe workloads on a common basis. The benchmark design is based on simulation of real world traffic conditions by defining a set of requirements which best characterize the end-users.
The Web Services Business Process Execution Language WS-BPEL 2.0, standardized by the OASIS consortium, is today's most commonly used language for the specification of orchestrated Web services. WS-BPEL is an XML based language that is used to describe the flow of information and control between Web services. BPEL processes are executed in so called BPEL engines that are often implemented as plug-ins to common Web/Application servers as Tomcat. These engines provide a complete run-time environment for concurrent BPEL processes that are triggered by respectively trigger themselves distributed Web services. This presentation describes the interdependencies between the configuration of a BPEL engine and its performance, utilizing the open source engine ActiveBPEL.
The experimental results compare the performance of the ActiveBPEL engine across different chip/system architectures and also help understanding some of its limitations. The prototype implementation is based on the TTCN-3 language which is used to specify the behaviour of the benchmark tests. In the simulation scenarios we have used different workloads to check the performance and reliability of the BPEL engine depending on the number of parallel invocations and concurrent processes. In parallel we have used different numbers of the working threads and compared the performance of ActiveBPEL engine running on Tomcat.
Added by: Deleted user Last edited by: Deleted user