web services performance comes of age in websphere application server (was) version 5.0.2, but just as with more traditional j2ee applications, the performance of web services applications is largely determined by the design of the application and the database.
this article considers the application design factors unique to web services performance, including the most important: moving to was 5.0.2. i will examine the performance of was 5.0.2 web services and establish some best practices for optimizing web services performance on was 5.0.2.
before discussing specific recommendations, we first need to understand the major performance characteristics or pressure points of web services.
web services performance components or pressure points
there are three major pressure points to a web services engine. these three pressure points define the performance "physics" of web services. they are defined here and shown in context in figure 1:
parsing (input): on the receiving side of a web service, the incoming web services payload is parsed. there are two major performance components in parsing: (1) scanner, and (2) symbol or name identification.
deserialization (input): as the document is parsed the xml payload is deserialized or marshaled into business objects that will be presented to the web services as business object parameters. the web services provider, either a javabean provider or ejb provider, has no awareness of its participation in a web service.
serialization (output): on output the inser... 下一页