I decided around the 24th of November that the way I was testing speed on each blog was inaccurate.

Therefore as of November 24th, we changed the way we are testing each host.

Fixed Page Size Testing

Now we are testing using a fixed size HTML file with 10k worth of text, or exactly 10,251 bytes. This same document is put on all servers we are testing.

Previously we were testing using the blog’s main index page which changes every day for each host and therefore not controlled or set fixed size.

In the results we post for November, you will see a sharp drop in ping / page load response times because of this switch. Now that the test page is only 10k in size, the response times sharply improved for each host.

If you want a more accurate result set, compare each screenshot test times after the 24th of November.

Faster Ping Resolution

Now that we are using a fixed page size, we could increase the resolution to get better results with our uptime reports. Previously we were testing using a 5 minute resolution, now we are using a 1 minute resolution. This will give us a better uptime percentage reporting for each host. Now if a host is down for a few seconds, we won’t report it as a 5 minute outage but will report it as a 1 minute outage. Still not completely accurate but it is a huge improvement.

In December, all days will have these new methods in place and will be a much more accurate result set.