	Department of Engineering

University of Cambridge > Engineering Department > computing help

The WWW's history at CUED

Introduction

CUED's first WWW pages go back to the early days of the WWW. At least one group had a server running in 1993. Our first departmental home page appeared in March 1994. Since then much has changed regarding WWW usage at CUED. Some of the more significant milestones will be noted here, and snapshots of activity provided.

Development of Services

The department provides various types of support to web authors and users.

Servers - We began by running a webserver on one of our general purpose machines. We now have one dedicated central machine providing 4 servers, and 2 other machines that do some web-serving too. A dozen or so research groups run their own servers, though we also provide a server for groups to use.
Cache-proxy - Since 1998 we've run our own cache-proxy which in turn feeds from the University and national caches. This reduces network traffic and speeds up access to commonly used remote files.
Information - The computer staff put information online to assist those creating servers and writing pages (the University now provide training and documentation too), and liaise with university and national bodies on behalf of the department. The Web Strategy Committee co-ordinates departmental activity and provides a House Style (dating from 1997; revised in 2005).
We have in-house experience with perl, Mason, PHP and JavaScript.

Development of Web Usage

Web usage has extended across nearly all areas of departmental activity, though progress has been slower in some areas than in others.

Use in Teaching - Impact is growing. In 2003 about 1% of courses had online material (typically a copy of the handout). One main course (1B computing) uses the WWW to provide material and check student progress. By 2006 many more courses had material online, a few using Web-based programs. Individual staff members have experimented with WWWboards, but because of the person-power required we have no Web-based teaching material.
On the plus side we have an extensive online archive of scanned cribs and past exam papers.
Teaching-related admin - Extensive use is made of the WWW. Syllabuses are online, also project selection, module selection, and student feedback are all web-mediated.
Admin - Most of the main documents intended for external users is now online - Annual Reports, Job vacancies, phone numbers, press releases, and details on how to get to the department. More recently (from 2000), purchasing has been done through the WWW. Increasingly, these pages are database driven.
Research - Many groups have extensive sites. Research grant and Research Career information is online for local users.
Computing - Much help-desk information is online.
Information for Prospective students - Both undergraduate and postgraduate prospectuses have been given special WWW treatment.
Library - The library has had an information page for many years. In 1999(?) the catalogue was made available via the WWW and the library acquired its own server.

Themes

Central/distributed control

Both in terms of server management and page production the balance between central and devolved control has shifted. For many years, any user has been able to put material online without central intervention. There was a phase when researchers had the same freedom to run webservers, but now this is unpopular both with research groups and the computer group because of security issues. Instead, we now maintain a centralised server on which groups can set up sites.

As regards page production, there was some pump-priming but now all the teaching and research pages and many of the admin pages are looked after by non-computer-staff - many are looked after by secretaries rather than academics.

Internal/external viewpoints

Initially one page served as our home-page both for internal and external users. Now we have a main home page for outsiders looking in, and a local page aimed at internal users, but little of the information on the latter there has access restrictions.

A few pages (directories of personal information, exam results, research project details) are for internal use only. More commonly material (press-releases, etc) is aimed specifically at external users.

Access patterns

We have log summaries going back to 1996. Interpretation of them isn't straightforward - as other sites use caches and as our research groups leave the centralised services to run their own servers, the patterns of access alter. Also we're moving towards dynamic pages (using PHP or databases) which affects logs too. Here are some points

Access hasn't always risen each year. Taking November as a typical in-term month, access to our home page has been as follows

Year Accesses

1996 7911

1997 12408

1998 17597

1999 (lost logs)

2000 35351

2001 49106

2002 62084

2003 80194

2004 72035

2005 69455

The relative popularity of many pages changes through the academic year, but here's a fairly representative sample of some term-time statistics from our main server in February.

Page	2002	2006
Intranet home page	14333	24168
Main research page	5020	6752
Undergraduate admissions	3140	3042
Postgraduate admissions	2933	1536
Directory of People	2863	2053
Main Teaching Page	1292	5960
Jobs Page	1078	2465
News and Events	814	725

The help system continues to account for about 33% of the accesses to our central servers. The server has many more pages than the other servers do (13k files, most of them not written by us), the pages are older, and linked-to more than files on our other servers are. Some of the specialist pages are amongst the most popular. Here's a sample of some statistics from February (sed is a little-used unix utility)

Page 2002 2006

Main LaTeX page 11056 14697

Main help page 3640 1520

Main C++ page 2286 1818

sed page 2078 3168

A graph of activity from June 2000 to November 2003 gives an indication of seasonal fluctuations.
Our central servers perform different tasks which in turn influences how much local use they get. The differences are striking. March 2002 is used in this first sample

Machine Purpose Requests Bytes sent % access from within Univ

www Central Server 1M 10G 65%

www-h Help System 0.7M 5G 3%

www2 Student Server 0.3M 6G 18%

www-g Group Server 0.3M 7G 48%

One would expect our central server to have many local accesses (it hosts the default home-page of our browsers). It also hosts scanned images which account for about 60% of the 'bytes sent'. In February 2006 (termtime) with more movies online, the stats were as follows

Machine Purpose Requests Bytes sent % access from within Univ

www Central Server 2.1M 71G 78%

www-h Help System 1M 8.2G 5%

www2 Student Server 0.5M 29G 18%

www-g Group Server 0.8M 20G 12%

| computing help |

Year	Accesses
1996	7911
1997	12408
1998	17597
1999	(lost logs)
2000	35351
2001	49106
2002	62084
2003	80194
2004	72035
2005	69455

Machine	Purpose	Requests	Bytes sent	% access from within Univ
www	Central Server	1M	10G	65%
www-h	Help System	0.7M	5G	3%
www2	Student Server	0.3M	6G	18%
www-g	Group Server	0.3M	7G	48%

Page	2002	2006
Main LaTeX page	11056	14697
Main help page	3640	1520
Main C++ page	2286	1818
sed page	2078	3168