Browse Prior Art Database

Shopping cart style aggregatory GETs for newspapers and magazines over the Web to efficiently use HTTP

IP.com Disclosure Number: IPCOM000012976D
Original Publication Date: 2000-Mar-01
Included in the Prior Art Database: 2003-Jun-11
Document File: 2 page(s) / 31K

Publishing Venue

IBM

Abstract

Shopping cart style aggregatory GETs for newspapers and magazines over the Web to efficiently use HTTP Disclosed is a system, method and program to improve the performance of websites of magazines and newspapers by aggregating the HTTP GETs from clients accessing the sites. Since many requests are conveyed by a client Web browser in one step, the load on the server is reduced. The solution is necessary because the design of the World Wide Web and the Internet has certain structural design problems. The problems include: 1. Time taken over the public Internet to convey data from Web Client to Web Server. 2. Overloading of Web servers by repeated tiny GET requests from clients for small amounts of content. When newspapers, magazines etc. are read over the Web, there are frequent small requests from the client. For example, while reading the business page of a newspaper there may be about 50 links, each link pointing to a 2K HTML document. A typical reader may click about 10 of these links to read the articles. Hence there are 10 GET requests to the newspaper and each time approximately 2K of data is transferred. Unfortunately because of slow-start and other problems with the TCP/IP protocol, sending 10 different times 2K of data is far less efficient that sending 20K of data once over the TCP/IP network. Furthermore, the load on the Web server for 10 GET requests is considerably more than if these 10 GET requests could be aggregated into one request. In fact some have said that the stateless HTTP protocol is an abuse of the TCP/IP protocol which is designed primarily for long sessions as in ftp or telnet.

This text was extracted from a PDF file.
This is the abbreviated version, containing approximately 51% of the total text.

Page 1 of 2

Shopping cart style aggregatory GETs for newspapers and magazines over the

Web to efficiently use HTTP

Disclosed is a system, method and program to improve the performance of
websites of magazines and newspapers by aggregating the HTTP GETs from clients
accessing the sites. Since many requests are conveyed by a client Web browser
in one step, the load on the server is reduced. The solution is necessary
because the design of the World Wide Web and the Internet has certain
structural design problems. The problems include:
1. Time taken over the public Internet to convey
data from Web Client to Web Server.
2. Overloading of Web servers by repeated tiny GET
requests from clients for small amounts of content.

When newspapers, magazines etc. are read over the
Web, there are frequent small requests from the
client. For example, while reading the business
page of a newspaper there may be about 50 links,
each link pointing to a 2K HTML document. A typical
reader may click about 10 of these links to read
the articles. Hence there are 10 GET requests to
the newspaper and each time approximately 2K of
data is transferred. Unfortunately because of
slow-start and other problems with the TCP/IP
protocol, sending 10 different times 2K of data is
far less efficient that sending 20K of data once
over the TCP/IP network. Furthermore, the load on
the Web server for 10 GET requests is considerably
more than if these 10 GET requests could be
aggregated into one request. In fact some have said
that the stateless HTTP protocol is an abuse of the
TCP/IP protocol which is designed primarily for
long sessions as in ftp or...