Browse Prior Art Database

Cache Management for WEB Sites with High Volume/Volatile Contents

IP.com Disclosure Number: IPCOM000015374D
Original Publication Date: 2002-Jan-27
Included in the Prior Art Database: 2003-Jun-20
Document File: 2 page(s) / 79K

Publishing Venue

IBM

Abstract

Disclosed is an application for managing very dynamic cache for a WEB site that have to publish the more recent information available in the system with a response time lower than 2 seconds for at least 90% of the requests. The cache should be shared among all the Web servers of the site in order to provide a consistent view of the system and to optimise the available resources.

This text was extracted from a PDF file.
At least one non-text object (such as an image or picture) has been suppressed.
This is the abbreviated version, containing approximately 54% of the total text.

Page 1 of 2

Cache Management for WEB Sites with High Volume/Volatile Contents

Disclosed is an application for managing very dynamic cache for a WEB site that have to publish the more recent information available in the system with a response time lower than 2 seconds for at least 90% of the requests. The cache should be shared among all the Web servers of the site in order to provide a consistent view of the system and to optimise the available resources.

This application was designed for The Olympic Intranet that had to manage a big amount of information with data that was changing very often (data were consolidated in different HTML pages) and with a very strong requirement for response times. Standard WEB Server cache procedures were not able to accomplish these requirements, and therefore a new cache management procedure was defined.

The cache was implemented by an AIX file system with the HTML pages stored in files. The different states of a page were coded in its file modes. Pages in the cache could have the following states:
valid: page ready to be served
expired: page with old information
locked: page being generated All the file attributes updates were done using a mutual exclusion mechanism, to avoid concurrency problems between the different tasks that were checking and updating them.

  Applications generating pages for the Olympic Intranet had specific processes to fill up the cache each time they received new data (see description of the pregeneration of HTML pages for high performance) or to expire the existing old pages in the cache.

The HTML cache was managed by two modules:

Page Selector. An extension of the Domino Go Webserver's base functions built using its

application interface (GWAPI). When an HTTP request arrived at the WEB Server this

high performance mechanism translated the PID (Page Identifier) into its corresponding

filename in the cache filesytem and directly returned the page if it was available.

Page Control. If the page was not ready in the cache f...