| |
Click here for full text:
Web Transaction Analysis and Optimization (TAO)
Garg, Pankaj K.; Hao, Ming; Santos, Cipriano; Tang, Hsiu-Khuern; Zhang, Alex
HPL-2002-45
Keyword(s): software performance monitoring; Web performance tuning; application response monitoring; software performance modeling
Abstract: Web user interactions start at a client (a browser on an access device) and flow through several software components connected via digital networks. While the business world is fascinated with the range of possibilities for such E-transactions, the underlying technology opens up several new challenges for managing the performance, availability, and reliability of the enabling software components. Several studies have shown that poor response times for Web user interactions can significantly impact the business values--slow Web transactions can have strong implications on brand-name, customer acquisition and loyalty, and purchasing behavior of customers. In the TAO project, we are developing metrics, models, and infrastructure to effectively manage the performance of Web applications. We use WebMon, a novel instrumentation tool to obtain transaction-level data from web interactions, from end-user and system component perspectives. Our analysis techniques help determine important segments of users and their web interactions. The analysis is embedded in visualization and optimization modules, enabling efficient reporting for system and business administrators, and automated resource scheduling and planning. In this paper we present an overview of the TAO project, and highlight some of its novel aspects, e.g., use of pixel-bar charts, web request classification, and integrated demand and capacity planning.
20 Pages
Back to Index
|
printable version
