The server for opennms.org has been around for almost five years now. It’s nothing fancy, but lately we’ve been running out of room on the disk drives so we went out and bought a new (refurb) server: lots of disk, lots of RAM, and two fast processors.
Unfortunately, I didn’t get the website moved fast enough.
The opennms.org site uses Mediawiki, which is written in PHP and had a MySQL back end database. It’s not the most efficient site in the world (although we have used various tricks to make it faster) but it has served us well. The usual load average is around 0.10 to 0.20, with occasional peaks when the nightly tape dump runs or someone is doing a lot of mail synchronization.
This week has been a big press week for OpenNMS. SearchNetworking.com issued its annual user survey and OpenNMS was awarded Gold in the network management platform category. This was thrilling for the team for a variety of reasons. First, this was a user survey. As an open source project we don’t really know if anyone is using our work. We have a wall of postcards but we don’t get many, and we have a large number of downloads but we can’t really tell if they ever get installed much less used. The fact that it was end users that honored us with this award means that someone, somewhere must find it worthwhile to use it.
Second, this was a survey of all tools, not just open source applications. People often want to associate open source with “cheap” or “bargain” and it is really the wrong way to look at it. OpenNMS can be better to use simply because it is a better tool for the job. When deployments are measured in man-hours, an OpenNMS deployment can take a lot less time than a comparable OpenView or Tivoli deployment. That’s where the real cost savings can be had, although the fact that there are no licenses fees involved is just a bonus. (grin)
Finally, it was OpenView and Tivoli that OpenNMS beat out for the Gold. We have always positioned the product against these platforms. Having used those applications in the past, I think OpenNMS is much easier to use and deploy. Many new users are overwhelmed by the sheer amount of flexibility OpenNMS offers, and it can scare some of them off. But OpenNMS was never designed to compete against tools like What’s Up or Nagios, which many find easier to install. However, if OpenNMS can monitor 50,000 devices at one site, it can really scream on a network with 500.
Another bit of press for OpenNMS came from Network World. I was in a face-off where the question was “Can You Trust Your Network to Open Source”. I, of course, took the “yes” position against Roger Greene from Ipswitch (makers of the aforementioned What’s Up product). It was easy to write the piece but much harder to do the podcast. See, What’s Up isn’t designed to be an enterprise-grade management platform. It is built to be a real easy, pointy-clicky application that small to medium businesses can use to get a view into the status of their network. For many, the time spent installing What’s Up may be less than working with, say, a Nagios installation. Note I am not saying “for all”, but I’m all about “if it works, use it” and if What’s Up works for you, then great. I liked talking with Roger and could see his point, although some others had to point out the flaws in his arguments a little more strongly. (grin)
But I do stand by my comments on the podcast that it’s ease of use makes it a commodity. And it is priced as such – a basic install of What’s Up is much less than an OpenNMS support agreement. But for those who need a high end solution for either complex systems or large networks, or simply a system that allows for easy customization, it’s going to require something like OpenNMS.
All is well and good until Monday afternoon when my e-mail stopped working. I tried to SSH into the box and couldn’t reach it, so I opened a support ticket and tried to figure out what happened.
What happened is that we were on the front page of Slashdot, and it was spanking our little five year old white box server. This is the first reply I got from my provider:
At console, your server was very slow, and very hard to work with; I was unable to determine why SSH failed to respond to remote connections. Your server is under a lot of stress, the last load average I was able to get:
load average: 162.58, 171.32, 149.41
I was finally able to get them to stop httpd which freed me up to work on the box, and I quickly moved the site to the new server. I upgraded to MediaWiki 1.9.3 and we installed squid front end for acceleration, and things look much better now (although it could be due to the fact that we’re not getting pounded like we were yesterday as much as the move to a new server).
To anyone who tried to visit us and failed, I apologize. And when you actually get to our sites, let me apologize again, because they aren’t the prettiest in the world. As we say on opennms.com: Professional Software, Amateur Marketing.