Research objectives and content
Given the success of the Web and the Internet, this research aims at investigating techniques to master the explosive growth of the Internet usage in order to (i) improve the response times, to (ii) support the introduction of new types of applications, and to (iii) assure scalability for very large user populations. The key idea is to use reliable multicast transmission for efficient replication and distribution of the information from the distant I servers closer to, or all the way to the clients. The benefits are better load distribution among the servers and a very efficient use of the available communications bandwidth. Two different scenarios for the use of reliable multicast to support Web caching and contents distribution will be studied, the necessary protocols will be designed, and a prototype will be implemented. The research will be in close collaboration and in extension of a larger ongoing project of the host institution that aims at designing and evaluating new multicast protocols for both, the reliable transfer of data and the timely delivery of audio/video data.
Training content (objective, benefit and expected impact)
Application scenarios and protocol mechanisms of the World Wide Web will be analysed to identify shortcomings of the existing paradigms for information distribution using the Web. Using the know-how for reliable multicasting of the host institute, new protocols for improving the performance, scalability and utility of the Web will be designed. The use of custom libraries for simulation developed at the host institute will be highly beneficial in quantifying the influence of the protocol parameters particularly for large groups and in cases where commercially available simulation tools would fail.
Links with industry / industrial relevance (22)
The work carried out will be an essential building block of any communications infrastructure, both in the Intranet and the Internet. The high speed network group of Prof. Biersack is currently working with Industry (Siemens Munich, CNET France) on the issues of reliable multipoint transmission. From this collaboration we expect valuable input and guidance for the research proposed here on 'Web caching and contents distribution' as well as a cross fertilisation. The contacts with the European branch of the WWW consortium that is located in proximity of INRIA Sophia Antipolis will be used to provide input and feedback for the evolution of the web.