Optimization approaches for accelerating dynamic content in E-business

Abstract

Information systems are increasingly becoming more interactive and dynamic in nature. Consider, for instance. the Internet, where web sites not only allow site visitors to purchase items and conduct banking online, but also serve pages that are personalized for individual site visitors. Such functionality is enabled by dynamic content generation technologies. In this paradigm, a user request is sent to a content server, which runs a program. This program accesses a set of data sources, creates a user deliverable object, and then returns this object to the user. In the context of the Internet, for example, a user may request the sports page in a news site. In this case, the server at the news site executes a program, which generates the user deliverable object-an HTML document corresponding to the requested sports page. There has been a significant amount of work on caching in the context of the Internet. However, most of this work is focused on caching rich content (e.g., images, multimedia objects) or caching at coarse granularities, such as HTML pages. A severe drawback of such coarse granularity caching is that the potential for reuse is often very limited. For instance, even though there may be significant reusability in the lower level objects on a page, the top level object (e.g., HTML document) is often unique (e.g., as in the case of a personalized page). In this paper, we address this problem by proposing a cost/benefit analytical model framework and a set of optimization techniques, which together will help identify the optimal set of cacheable objects. Our work will extend the application of caching in the context of the Internet to a much wider variety of objects. Our proposed approach entails the following steps: We will develop an object data model to represent dynamically generated web content; Based on this object model, we will propose a cost/benefit analytical framework, which can be used to measure the costs and benefits associated with caching objects; Based on this object model and cost/benefit framework, we will propose a number of optimization approaches to extract the optimal set of cacheable objects; Finally, we will present the results of an extensive performance evaluation of the proposed optimization techniques with regard to both accuracy and efficiency.