Approved for Public Release

This paper proposes a method for improving the performance of Web servers servicing static HTTP requests. The idea is to give preference to those requests which are quick, or have small remain-ing processing requirements, in accordance with the SRPT (Shortest-Remaining-Processing-Time) scheduling policy. The implementation is at the kernel level and involves controlling the order in which socket buffers are drained into the network. Experiments use the Linux operating system and the Rash web server. All experiments are repeated under a range of server loads and under both trace-based workloads and those generated by a Web workload generator. Results indicate that SRPT-based scheduling of connections yields significant reductions in mean response time, mean slowdown, and variance in response time at the Web server. Most significantly, and counter to intuition, the large requests are only negligibly penalized (or not at all penalized) as a result of SRPT-based scheduling