It is possible to derive the optimum pre-fetching sequence of the nodes that will minimize the expected cumulative read-time lag for a given network bandwidth for a given graph pattern.

Proactive Web Caching and Prefetch

This research is engaged in analytical model based techniques for information prefetching and caching.

 

Caching and prefetching are the two principal techniques, known for improving responsiveness for systems that involve data communication. A number of recent innovative but empirical schemes in web systems report efficacy in the range of 30-60%. Interestingly this is quite low as opposed to the +90% hit rate easily achievable in hardware.

 

The variability of parameters faced in the Internet is quite high, and it makes internet caching/prefetching a harder problem than it’s hardware counterpart. In the web, there is no short-term iterative construct such as 'while' of 'for' loops in the causal chain above.

 

The degree of branching in web is very high. Consequently in web-systems caching/prefetching often results in excessive data transfer.

>> WW: Javed I. Khan and Qingping Tao, Exploiting Webspace Organization for Accelerating Web Prefetching , International Journal of Web Intelligence and Agent System, IOS Press, Netherlands, vol.3, No.2, 2005, pp117-130, (GS(3)). [WIC] [KhTa05]

 

>> WW: Javed I. Khan and Qingping Tao, Exploiting Webspace Organization for Accelerating Web Prefetching , IEEE/WIC International Conference on Web Intelligence, WI'03, October 2003, Halifax, Canada, pp.89-95. [KhTa03b]

 

>> WW: Javed I. Khan and Qingping Tao, Webspace Surfing Patterns and Their Impact on Web Prefetching, Proceedings of the IEEE 2003 International Conference on Cyberworlds, CYBERWORLDS2003, Singapore, December 2003, pp478-485. [KhTa03a]

 

>> WW: Javed I. Khan and Qingping Tao, Partial Prefetch for Faster Surfing in Composite Hypermedia, 3rd USENIX Symposium on Internet Technologies and Systems, USITS 2001, San Francisco, March 2001, pp13-24. [KhTa01], (GS(17)).

Selected Publications

More Publications >>

In this research, armed with the new insight given by the formal tools we are focusing in creating novel optimization techniques for internet prefetch. We have elicited one of the first techniques rooted in formal analysis-- with guaranteed optimum ranking of the prefetch paths. We have shown that instead of full prefetch proposed in the first generation web systems— a new technique of ranked partial prefetch can dramatically reduce the waste in prefetch, without compromising the system’s responsiveness gained in the first place.

 

For a given bandwidth partial prefetch can stage more potential links, resulting in dramatic speedup in systems responsiveness.

 

This research has been sponsored by Ohio Board of Regent’s Research Challenge Grant.

Publications

Distributions

Demo

Technical Reports

All Projects

 

Page last updated February, 2008, Medianet Laboratories.