Question for all Tech Support regarding customers

[cwscribner]

Hi all.

Is there a list of current implementations and their sizes anywhere? We just want to know if we're "breaking new ground" with the way we're implementing Nagios or if there are other Nagios users covering the same things and encountering the same issues.

[mguthrie]

Unfortunately we don't have a list like this. A lot of customers have environments that are constantly changing. I've helped customers with monitoring environments ranging from 50 to several thousand hosts. However, you can post a profile of your environment along with some of the challenges you're coming across and there may be some users with ideas. We've also got our world conference coming up, and there are people who will be presenting there who have pushed both Nagios Core and XI far beyond anything we've replicated in-house (there's a guy there who uses Nagios for 1.4 million services), so hopefully that will give us some new ideas. We're also happy to discuss new ideas or ways to break new ground with our software, so if there are ideas or things you'd like to try, by all means let us know.

[cwscribner]

My main reason for asking this is the size of our implementation. We have 3262 hosts and 698 services and we're only 25% of the way through the project. So in the end we'll roughly be supporting 12-14k devices. As the lead (read: only) tech on the project, I'd like to see what other customers are doing when dealing with large implementations. How many servers, configuration, special Nagios modifications, etc. Is there perhaps a sub-forum where active customers can discuss these topics?

[mguthrie]

There isn't currently. Right now for both XI customers and the support team this is the most visible place. It's definitely a goal of ours to be able to support larger monitoring environments with XI, so we'd definitely be up for discussing ideas and challenges are you move forward with it. Something that might be worth looking into as the environment grows past 8-10k checks would be DNX. It works like a load balancer to distribute your Nagios checks across multiple slave machines. Our best estimates would guess that a single server-class machine can probably handle around 8-10k checks with a 5-10mn check interval. This number decreases if a lot of the checks are either SNMP or checking VMWare. If it's higher than that then you might want to either split it between two XI instances and use Fusion for a high-level overview, or use the load balancer model with DNX.
http://library.nagios.com/library/produ ... ith-nagios

The nagios users mailing list on sourceforge is followed by a lot of long-running Core users, some of which run pretty huge environments. They might also have some good ideas, but we've done our best to study and document stable methods for larger environments.