Database of Munin Installations

These past months have seen a significant number of new users on the ML or on IRC having performance problems with Munin. There installation are much bigger than the several nodes that Munin was designed for at its beginning.

There are some well-known optimizations that can be done on a stock Munin, but the impact of each optimization isn't well known. That's quite a pity, since we really want Munin to scale well, and be able to handle even biggest installs, while retaining its core motto : "Be dead simple to install".

A great step forward has been to take CGI graphing to next level in the upcoming 2.0. Since graphing was very resource intensive, on-demand graphing was a performance no-brainer. Coupled to FastCGI to avoid startup times, the graphing is almost as fast as serving the precomputed static files for a reasonable amount of users.

This page is here to collect statistical data on existing munin installs. It will help us to see where to focus our efforts, and at the same time, give some clues to new users that could see what sort of hardware & software configuration to suit there needs.

We know that the data we are speaking here is usually very sensitive for companies, so these are only anonymous data. Please send a private email to steve.schnepp _at_ gmail.com if you want to be added here.

Version	# nodes	# RRD files	RRD FS	# Cores	RAM	running time (s)
2.0b1	1	100	EXT4	1	0.5	40