Ruth E. Willenborg

Simplifying Service Deployment with Virtual Appliances

As IT services become more powerful and complex, service deployment gets more difficult and expensive. Service deployment, the process of making a service ready for use, often includes deploying multiple, interrelated software components into heterogeneous environments. Different technologies and tools try to address these complexities by describing the environments, abstracting the dependencies, and automating the process. Virtual appliances, a set of virtual machines including optimized operating systems, pre-built, pre-configured, ready-to-run applications and embedded appliance specific components, are emerging as a breakthrough technology to solve the complexities of service deployment. Virtual appliances provide a simple, unified and easy to use interface for service deployment by encapsulating entire custom environments, and resolving the execution policy constraints and inter-dependencies through pre-installing the software applications. The motivation of this paper is to prove virtual appliances offer a better service deployment mechanism. We start with an easy to understand model to describe the complexity of service deployment and introduce the architecture of a virtual appliance. We then analyze the deployment process of using traditional deployment mechanisms, and quantitatively and qualitatively compare the deployment time, operations and parameters of the traditional approach with the use of virtual appliances. The results show virtual appliances offer significant advantages for service deployment by making the deployment process much simpler and easier, even for the deployment of advanced enterprise services.

Designing WebSphere application server for performance: an evolutionary approach

Performance and scalability are critical elements for a successful Web site and therefore, are fundamental design criteria for the IBM WebSphere® platform. This paper focuses on the evolution of IBM WebSphere Application Server performance and scalability features, the improvements that were achieved, and directions for future work. There are three design principles underlying the success of WebSphere: (1) optimize the design for the predominant (no-failure) case, (2) make finite resources appear infinite, and (3) minimize the number of interprocess calls. We illustrate the application of these design principles in key areas, including workload management, Hypertext Transfer Protocol (HTTP) session management, back-end connection management, session data and content caching, and Enterprise JavaBean™ (EJB™) design and deployment patterns.