Summary: Improving HTTP Latency
Venkata N. Padmanabhan (University of California ญญ Berkeley)
Jeffrey C. Mogul (Digital Equipment Corporation Western Research Laboratory)
The HTTP protocol, as currently used in the World Wide Web, uses a separate TCP connection for each
file requested. This adds significant and unnecessary overhead, especially in the number of network round
trips required. We analyze the costs of this approach and propose simple modifications to HTTP that, while
interoperating with unmodified implementations, avoid the unnecessary network costs. We implemented our
modifications, and our measurements show that they dramatically reduce latencies.
People use the World Wide Web because it gives quick and easy access to a tremendous variety of information in
remote locations. Users do not like to wait for their results; they tend to avoid or complain about Web pages that
take a long time to retrieve. That is, users care about Web latency.
Perceived latency comes from several sources. Web servers can take a long time to process a request, especially
if they are overloaded or have slow disks. Web clients can add delay if they do not quickly parse the retrieved data
and display it for the user. Latency caused by client or server slowness, however, can in principle be solved simply
by buying a faster computer, or faster disks, or more memory.
Web retrieval delay also depends on network latency. The Web is useful precisely because it provides remote
access, and transmission of data across a distance takes time. Some of this delay depends on bandwidth; one cannot
retrieve a 1 Mbyte file across a 1 Mbit/sec link in less than 8 seconds. You can in principle reduce this time by