D. Sarkar (USA), U.K. Sarkar (India), and G. Peng (USA)
Cache servers, Traffic engineering, Document retrieval latency, Hierarchical caching network, Link bandwidth, World Wide Web
Document retrieval latency is a major problem in the Inter net. Caching of documents at servers close to the clients reduces the latency and many such caching schemes are known. But, the impact of a caching scheme on the band width requirement of individual links of a caching net work is little known in the literature. Knowledge of es timated bandwidth requirements of links is useful for the study of latency and also for the management of total avail able bandwidth, a limited network resource, for allocation across the links of a caching network. This work formulates a graph-based framework and methodology for estimating the relative bandwidth requirement of links in a hierarchi cal caching network. It uses the fact that an existing latency will not worsen if the bandwidths of non-bottleneck links are reduced until those become bottlenecks.
Important Links:
Go Back
