This is actually tracking how many unique IPs accessed bugzilla-update.xml over a 30-day period, each day. So, for example, if we’re talking about August 1, 2010, then the number that is displayed for that day is how many unique IPs accessed bugzilla-update.xml between July 1 and August 1.
Not all Bugzilla installations check bugzilla-update.xml. It’s only checked if an administrator views the home page on their Bugzilla installation, and only Bugzilla 3.0 or above do this. Also, “unique IPs” does not directly translate to unique installations.
This should still give us a fairly good idea of how our installation count changes over time, though. It’s more useful to look at the trends of the graph than the actual exact numbers.
The real number of active installations at any given time is likely higher than the numbers here. Most servers don’t change their external IPs, so we wouldn’t be getting too many Bugzilla installations whose IPs change in a 30-day period. On the other hand, there are probably multiple installations behind NAT at various organizations, many people are running ancient installations before 3.0, lots of people probably turn off the update checker, and there are a lot of Bugzilla installations that operate in environments where they can’t access the remote bugzilla-update.xml.
NOTE: We lost active installation data from 9. May to 8. Sep 2012 (4 months) due to a bug in the generation script after moving the system to a new server. Sorry!