Harvey Gunther – IBM WebSphere Product Performance Group

Fixing J2EE Application Performance – A Practical Example

Using a hypothetical application as a case study, Harvey will examine how to fix the performance problems of a poorly written and incorrectly tuned runtime for an application utilizing Servlets, JSPs and EJBs. Step by step, he will show how to tune and modify this application to realize its performance potential. After each step, the performance impact of that step will be established and compared. In this session, Harvey will also examine and compare the alternatives of fixing a badly performing application or adding additional resources, including application servers and hardware.

The following will be covered:

• J2EE best application development practices for performance and scalability. We will examine and discuss illustrative code snippets.
• Performance methodology for examining application performance. This will include benchmarking and examination of the three key performance metrics, throughput, CPU utilization and response time.
• Diagnostic techniques including Java thread dump analysis and application code profiling.

Application Performance Considerations for Web Services and J2EE Components

In this session, Harvey will examine the performance characteristics of Web Services using J2EE components. He will consider a B2B scenario involving a manufacturer/client application conducting business with a supplier through Web services. The manufacturer/client application uses UDDI to locate multiple published suppliers of a particular part. All business between the manufacturer and the supplier uses SOAP.

The following specific issues will be considered:

• Performance Methodology for distributed applications using Web Services
• Best performance practices for SOAP, UDDI and WSDL
• Performance characteristics of web services, including the impact of the XML parser
• Performance impact of the size and complexity of the underlying XML payload
• Performance contrast of the underlying network transports including HTTP and HTTPS
• Performance techniques including thread dump analysis and profiling
• Diagnostic methodology for distributed applications using Web Services. Specifically, this includes tracing and examining the underlying XML payloads and debugging "SOAP FAULTS"

Harvey Gunther is a Senior Performance Analyst with the IBM WebSphere Product Performance Group. Currently, Mr. Gunther is focusing on web services performance on WebSphere. Mr. Gunther also specializes in application performance issues and best practices for performance and scalability. Prior to working in WebSphere Performance, Mr. Gunther was a Senior Consultant with IBM's AIM Services organization. In this capacity, Mr. Gunther worked with customers on WebSphere performance issues and application design issues. Mr. Gunther started with IBM as a performance analyst in IMS Performance. Email: hgunther@us.ibm.com

Back to ...

dIon Gillard

On to ...

Peter Haggar