We've implemented a test RPC style web service using the following methods: 1. EJB handler with manually created deployment descriptor. 2. WSDL2Java (thus RPC provider) with auto-generated deployment descriptor.
Note that both web services implement identical operations in their implementation classes. We then did a performance comparison between the two and we discovered that the EJB handler based web service is a lot faster than the WSDL2Java generated web service. Can some please give me a reason for this disparity in performance between this two types of web services.
