Performance Metrics

We consider five aspects of web caching benefits: hit ratio, byte hit ratio, latency reduction, hop reduction, and weighted-hop reduction. By hit ratio, we mean the number of requests that hit in the proxy cache as a percentage of total requests. By byte hit ratio, we mean the number of bytes that hit in the proxy cache as the percentage of the total number of bytes requested. By latency reduction, we mean the percentage of the sum of downloading latency for the pages that hit in cache over the sum of all downloading latencies. Hop reduction and weighted-hop reduction are used to measure the effectiveness of the algorithm at reducing network costs, as explained below.

To investigate the regulatory role that can be played by proxy caches, we model the network cost associated with each document as "hops". The "hops" value can be the number of network hops traveled by a document, to model the case when the proxy tries to reduce the overall load on Internet routers, or it can be the monetary cost associated with fetching the document, to model the case when the proxy has to pay for documents traveling through different network carriers.

Associated with the "hop" value are two metrics: hop reduction and weighted-hop reduction. Hop reduction is the ratio between the total number of the hops of cache hits and the total number of the hops of all accesses; weighted-hop reduction is the corresponding ratio for the total number of hops times "packet savings" on cache hits. A cache hit's packet saving is 2 + filesize/536, as an estimate of the actual number of network packets required if the request is a cache miss (1 packet for the request, 1 packet for the reply, and size=536 for extra data packets, assuming a 536-byte TCP segment size).