At the end of the experiment, the master process receives from clients the number of requests sent and the total number of bytes requested and from servers the total number of requests and bytes that they serviced. Clients also send the average latency and the total number of errors observed during the two phases of the experiment. An error is reported everytime a connection between client and proxy could not be established, because the pending connection queue was full. Then, the master process generates an overall performance report that contains the following data: average latency in the first and second phase, total number of errors observed during both phases, hit ratio and byte hit ratio. Proxy throughput, i.e., the number of requests that are handled in each time unit, can be estimated based on the average latencies and on the number of requests handled.