WWW94: DEC's WWW server, a case study
presented by russ jones, digital equipment corporation
introduction
Digital Equimpent Corporation (DEC) was one of the first large companies
that provide information to their customers, prospects and partners via
the web. the author presented the major goals of their approach,
the actual implementation plus the results and what was learned during
the process of setting up a WWW server.
goals
DEC wanted its server to be setup as fast as possible ("time to market").
since the basic information was already there as an FTP archive, all that
was needed was a way to present the data in WWW-style.
the FTP archive is divided into directories. each document is available
as an ASCII textfile and/or as a postscript file. the later is often
also available as a ZIP file and/or in other compressed formats. plus
there is an abstract for each document.
DEC also wanted to establish several indexes so the readers could find a
document in many different ways. users shall be able to navigate by:
- alphabetical document title
- document type (technical overview, performance report, etc.)
- document subject
- chronological order (actual reverse chronological order)
- random word search
design issues
DEC wanted to give their documents a unified look. a basic style guide was
set up. beside other rules, each page should have:
- the digital logo on top of it
- a meaningful title
- a creation date
- the initials of the author
while experimenting with graphics, it has become clear the images must be kept
small and rare. too many or too large images lead to unacceptable response
time because a huge amount of data has to be transferred over the network
which in turn distracts users from reading such pages.
implementation
because most of the information was already in place - as mentioned above -
the basic work was to generate HTML documents, that would include the
text from the abstract and links to the related files. beside that, the
various indexes have to be built. both operation were automated.
building the indexes was rather easy. the procedure runs every night and
rebuilds the indexes to include new documents.
also most of the HTML documents could be generated automatically from the
abstracts. however some restrictions became apparent. by definition, HTML
leaves the formatting of the document to the client. currently HTML does
not support special formats like tables, TABs, etc. so in some cases it
was necessary to convert the document manually or the original ASCII
text was wrapped with the "preformated tag" which allows to preserve to
format of a text.
results and lessons learned
it took DEC five months to setup a WWW server and to prepare the data.
when it went on-line on october 1993, there were 1500+ HTML documents
containing about 32'000 links.
one of the major problems encountered during the first few months was the
fact, that they used URLs with absolute addresses. therefore it was not
possible to move documents around or even to another system. all URLs that
referred to another HTML document where translated into relative addresses.
i have no link information for this paper on the web, but you may want to
browse through
digital's WWW server.
2nd_day_dec /
13-jun-94 (ra) /
reto ambühler
!!! Dieses Dokument stammt aus dem
ETH Web-Archiv und wird nicht mehr gepflegt !!!
!!! This document is stored in the
ETH Web archive and is no longer maintained !!!