Date of Award

2010

Degree Type

Thesis

Department

Computer and Information Science

First Advisor

Arndt, Timothy

Subject Headings

Ajax (Web site development technology), Web site development, AJAX, Internet, Concurrent AJAX, Web Performance, Web

Abstract

The increasing use of the Internet and demand for real-time information has increased the amount of dynamic content generated residing in more complex distributed environments. The performance of delivering these web pages has been improved through more traditional techniques such as caching and newer techniques such as pre-fetching. In this research, we explore the dynamic partitioning of web page content using concurrent AJAX requests to improve web page delivery performance for resource intensive synchronous web content. The focus is more on enterprise web applications that exist in an environment such that a page's data and processing is not local to one web server, rather requests are made from the page to other systems such as database, web services, and legacy systems. From these types of environments, the dynamic partitioning method can make the most performance gains by allowing the web server to run requests for partitions of a page in parallel while other systems return requested data. This differentiates from traditional uses of AJAX where traditionally AJAX is used for a richer user experience making a web application appear to be a desktop application on the user's machine. Often these AJAX requests are also initiated by a user action such as a mouse click, key press, or used to check the server periodically for updates. In this research we studied the performance of a manually partitioned page and built a dynamic parser to perform dynamic partitioning and analyzed the performance results of two types of applications, one where most processing is local and another where processing is dependent on other systems such as database, web services and legacy systems. The results presented show that there are definite performance gains in using a partitioning scheme in a web page to deliver the web page faster to the user

COinS