Hosting lab test:
where we compare web hosting services and find the best
home for your site
By Jon Smith
Each month Internet Magazine will compare web
hosting companies to find the best Web hosting
out there. With our compare web hosting
results, you can make sure your site's getting the best
Here are the results of the compare web hosting
work: We have a new leader in our table this month --
Entanet. This Telford-based host has been performing
consistently for a long period of time, and finally
got its break after a slip-up by last month's runaway
leader lnweb Networks.
When you compare web hosting services
you will find suprises. To compare web hosting
companies you first need to know what your site needs.
There were a handful of excellent improvements we found
in our compare web hosting search.
Here are more results from our compare web hosting
efforts: Another host with Telford connections, Wyre
Forest Computer Solutions, is up dramatically from 34th
to fourth. Christian Technology climbs 22 places from
28th to sixth, Imagine Media is in 12th from 39th 100
Percent IT has improved from 29th to 14th, and NetWeaver
is looking good in 18th up from 26th.
More UK results from our compare web hosting
search: Netscalibur won't be happy to have fallen from
eighth to 23rd despite being one of only four hosts
to post 100 percent availability, and Routo and Harbourweb
have fallen out of the top 20, but all in all, this
month's table reflects a period of consistency in the
UK hosting market.
How it all works
Our top throughput figures can exceed 3,000kbps, yet
modems generally manage throughputs of not much more
than 5kbps. The principle is that if you get 3,000kbps
on Zeus' high bandwidth connection, it's the equivalent
of 600 modem connections downloading at 5kbps.
Rankings are based on transfer speeds--how quickly files
are downloaded from the host server to the test machines.
Zeus uses five test hosts in the UK and five in the
US. While every site is tested by both the UK and US
test hosts, the US testing generates better results
for machines that are located in the US. The test files
used are between 4,096 and 6,144 bytes. The only requirement
is that they are static files of the right length, so
a text file with 5,000 '*'s in it would do. Images are
generally used because it makes life easier for the
sites being tested to find an image of the right size.
Many of the page files we test are the hosts' own home
pages. Although not all clients will achieve the same
levels of performance, the thinking is that if a host
can't get its own pages to download rapidly, it's unlikely
to manage any better with customers' pages.
Availability figures are based on availability of the
Web server itself, not server uptime. Our systems attempt
to contact the host every 15 minutes, 24 hours a day.
The tests ran from 26 May to 24 June 2002. For more
information, visit Zeus Technology's real-time rolling
Web server tests at www.webperf.net.
This month's best-performing hosting companies: