USENIX Symposium on Internet Technologies and Systems, 1997
HPP: HTML Macro-Preprocessing to Support Dynamic Document Caching
Fred Douglis, AT&T Labs - Research
Antonio Haro, Georgia Institute of Technology
Michael Rabinovich, AT&T Labs - Research
A number of techniques are available for reducing latency and
bandwidth requirements for resources on the World Wide Web, including
caching, compression, and delta-encoding .
These approaches are limited: much
data on the Web is dynamic, for which traditional caching is of
limited use, and delta-encoding requires both a common version base against which to
apply a delta and the complete generation of the resource prior to
encoding it. In contrast to these approaches, we take an
application-specific view, in
which we separate the
static and dynamic portions of a resource. The static
portions (called the template) can then be cached,
with (presumably small) dynamic portions obtained on each access.
Our HTML extension, which we refer to as HPP (for HTML Pre-Processing) supports resources
that contain variable number of static and dynamic
elements, such as query responses.
Results with macro-encoding of query response resources from local CGI scripts and
two popular search engines
indicate that our approach
promises a substantial reduction
of network traffic, server load, and access latency
for dynamic documents. The size of network transfers using HPP are
comparable to delta-encoding (factors of 2-8 smaller than the
original resource), while the data generated by content
providers is simpler, and the load on the end-servers is
- View the full text of this paper in
HTML form and
- If you need the latest Adobe Acrobat Reader, you can download it from Adobe's site.
- To become a USENIX Member, please see our Membership Information.