WWW: Beyond the Basics

25. Methods for Web Bandwidth and Response Time Improvement

25.1. Introduction

As of April 1995, the last month for which statistics are available, the Web has become the largest single component of traffic on the NSFNET backbone in both packets (21.4 percent) and bytes (26.5 percent) (Pitkow, 1996). The emergence of the Web has stretched the network technology to the point that "brown­outs" frequently occur on the Internet (a collection of networks). High network traffic typically causes these failures. These problems have lead people to start calling the acronym WWW as the World-Wide Wait.

This high traffic problem will continue to plague the Internet for the foreseeable feature, especially considering the emergence of new multimedia traffic (video and audio). Thus, one important consideration to the Web's current architects is how to reduce network traffic. Approaches to reducing traffic includes removing it entirely from the network and moving it from congested backbone networks to higher speed local networks.

The other common complaint is poor user response time. Response time is generally measured as the time between when a request is issued and the response is received. It is closely related to the time it takes to transfer a document. This metric is very important to the user since it provides a measure on how long it takes for a document to be received.

The most obvious way to improve Web performance is to obtain higher bandwidth connections and faster routers. To get the improvement, the upgrades must occur in many places. These are the connection end­points and all the intermediate networks. This is prohibitively expensive to do on a world­wide scale. Another way to obtain better Web performance is to reduce bandwidth consumption by intelligently sending data only when absolutely necessary. This is the primary solution that is discussed in this chapter.

There are three general intelligent bandwidth consumption methods to improve Web performance. These are protocol modifications, document caching, and alternative distribution methods. Protocol modifications have an immediate impact on performance but may result in significant infrastructure changes that users may not accept. Document caching is generally transparent to the users and Web document designers and is under active research. Alternative distribution schemes may help alleviate the problem but does not hold the same potential as the other methods.

[PREV][NEXT][UP][HOME][VT CS]

Copyright © 1996 David C. Lee, All Rights Reserved

David C. Lee <dlee@vt.edu>
Last modified: Mon Nov 18 15:22:36 EST 1996