Browse Prior Art Database

A Session Data Management Method for Web Applications

IP.com Disclosure Number: IPCOM000013893D
Original Publication Date: 2000-Aug-01
Included in the Prior Art Database: 2003-Jun-19

Publishing Venue

IBM

Abstract

Disclosed is a system for efficient session management of Web applications. In particular, it contributes both to the reduction of system memory usage and to ease of use during the navigation among transactional Web applications, which require consistent session management with comparatively large memory requirements in many cases. Accordingly, the disclosed algorithm enables scalable transactional systems using Web infrastructure. The basic idea consists of (1) watching the access history of URLs which the user visits by indirectly accessing external URLs, (2) defining hierarchical classes of URLs, which are used to select an appropriate session management strategy according to the currently accessed URL, and (3) storing the session data into different parts of memory with different life times which are determined by the session management strategy. When a user starts a session, some session information is stored in the server's main memory. In the processing, the user may access internal or external URL's outside of the session-related pages. The user may return to the session soon or much later or may not return at all. The likelihood of return can be calculated from the URL class of the currently accessed URL. According to this probability estimate, the session data is swapped out of the main memory to secondary or other complementary stores, which have larger capacity but require expensive costs to access the data. In addition, if the user does not return to the session within a determined expiration time, the swapped session data will be expired to release the resources, where the expiration time is also determined based on the probability of the session being resumed. Web Application Web Server