About LSAM Request Routing

Cache Request Routing

The goal of request routing is to reduce origin server load, bandwidth usage, and user delay, by forwarding HTTP requests to proxies that are likely to have cached copies of requested data. Finding cached copies in nearby proxy servers will, in general, achieve all of these objectives. The approach to achieving these advantages is to establish a system of caches and routing between them such that aggregate request streams are formed that can exploit commonly requested pages. This requires (1) a system that establishes a "mesh of caches" so that stream aggregation is possible, and (2) a routing algorithm that balances potential advantages of cache hits with the delays experienced by a series of cache misses. This routing system is being designed to both support and augment the primary LSAM mechanism for efficient distribution of related data files via multicast channels.

We believe that the low cache hit rates seen in standard hierarchical cache systems may be improved by adding the LSAM multicast distribution. Since LSAM multicast distribution will send "popular" files to points of interest ahead of time, this can reduce the number of requests flowing through the cache system and consequently will reduce the "turnover" rate in the cache.

Discussion and Details

More Information

Contact Steve Hotz for more information about LSAM Request Routing. Stephen Suryaputra is the GRA working on request routing.