A proxy server structure and its cache consistency mechanism at the network bottleneck

The response time, bandwidth usage and server load are three representative performance metrics that are used for proxy cache consistency mechanisms. However, the primary metric affecting the overall performance can be different depending on the situation. At a network bottleneck, such as the inter-continental backbone, the network bandwidth is limited. Also, the response time is a basic performance metric used for the proxy cache, and an unsatisfactory response time can be a reason why clients hesitate to use the proxy cache, which may potentially increase the network traffic. Hence, we should give higher priority to the response time and the bandwidth usage at the bottleneck, as compared to other metrics. In this paper, we suggest a proxy server structure at the network bottleneck which improves the response time and even results in a small amount of network traffic reduction at the bottleneck. Its cache consistency mechanism is based on periodic polling performed at the polling server located beyond the bottleneck. Our approach may increase the network bandwidth beyond the bottleneck and the load on the originating Web servers. However, the faster response time and the reduced network traffic at the bottleneck can compensate for them