Chapter 19 - Performance Analysis and Optimization Techniques

Abstract

Although service-oriented architecture (SOA) promotes the benefits of interoperability, reuse, and rapid development, poor performance will spell disaster. The SOA architect and service designer must include strong consideration for performance, capacity, and scale in the design process. There are several contributing factors to evaluate and optimize SOA and Web services, and each can significantly affect the overall performance of the SOA. Consideration should be given to the complexity of the internal consumer and service behavior, the ESB, security requirements, and design of the message. Code reviews including a view toward performance are strongly recommended. ESB is the primary SOA technology capability and component. If the ESB does not have sufficient capacity or it cannot scale to meet demand, latency will result. Selecting the ESB should include consideration for point in time performance requirements, as well as longer-term growth and trending. Accelerators are technologies that have been optimized for specialized processing such as XML validation and XML message transformation. Ensuring security of the SOA, service interactions, messages, and data are critical. There are many approaches to securing an SOA or Web service, and many will require additional or specialized processing such as encryption and decryption. The complexity and size of a message is possibly the most obvious and easily optimized area of an SOA. Validation of XML-encoded messages is another potential area of performance impact. Validation can help to ensure compliance with the service interface contract and also avoidance of data anomalies in the message.