Month: September 2007

Mont Blanc

Posted on by tarus

Ah, the joys of international travel in coach. Business class fares to Geneva from RDU are obscene, so I flew coach. RDU to LGW, eight hour layover, then LGW to GVA. My friend Alex picked me up and drove me to his house in France, where we stayed up until midnight, and after pastis, fondue, wine and a lot of good conversation I slept until 2pm the next day.

Since the languages of choice in Alex’s family are French and German, I spent the weekend totally lost, although I did play enough with my French Rosetta Stone lessons that I can say, with authority, “Un homme sur la table” (The man is on the table). I can’t tell you how useful that phrase has been, but you can guess.

On Sunday we went to Chamonix, to visit Mont Blanc, the highest peak in western Europe. You can take a couple of cable cars up to Aiguille du Midi, a point on the side of the mountain that is about 1000 meters from the summit.

Unfortunately, the clouds came in as we were ascending and only parted briefly for a view of the summit. Mont Blanc is taller than Mount Whitney, the tallest peak in the continental US, and twice as tall as the tallest mountain near me, Mount Mitchell. The second cable car ride was a little scary for me, and I don’t usually have problems with heights. It had something to do with being really close to the rock face during the final ascent.

Just as an example of how the Internet has changed things, we met up with two climbers who had walked up to the Aiguille. They were from the Czech Republic and when they saw Alex’s camera, they asked if he would take some pictures and e-mail them. None of us had a pen so Alex entered in the e-mail address on his phone.

E-mail, digital photos and text on phones are commonplace, but it was a little strange to be using them considering how little modern technology had been used to get us to that point on the mountain.

On Monday I went to work in Geneva. I am working with an old client of ours and as much as I like the Internet there is nothing like meeting face to face. OpenNMS 1.3.7 is really amazing, and I was finally able to address some issues that we had been struggling with via e-mail, once we had the chance to sit around a table and go over them. I’m working on the upgrade now, and I expect performance to be an order of magnitude better on this large installation. These days the network is so good that I was able to VNC to my systems at home, and while not snappy the performance useable – from a continent away. As we worked toward 1.8 it is like we’ve climbed our own mountain, and I can finally see the summit.

Although at times it seems obscured by clouds.

Off to Yurrip

Posted on by tarus

Well, I’m about 2 hours into an 8 hour layover at Gatwick, on my way to Geneva. I’m chillin’ in the BA lounge and trying to get caught up on work. I haven’t slept in about a day and there is still a long way to go before I’m in bed a friend’s house in the French Alps. Luckily he just came back from the States yesterday and should be suffering from jet lag too. He has three kids, but I think I can sleep through anything at this point.

To go on this trip I had to miss the last day and a half of training. It’s a lot of fun to run training out of the office. The students don’t get pulled away as much as they do when I train on-site, the class size is small so we get to talk a lot about real-world problems, and since we’re holding it at the office we can easily share the training load if, say, I have to get on a plane.


David Explains Distributed Monitoring

I taught the first three-day Basic course this week, and David is running the two day Advanced course. I kind of wish I could have sat in on it, because OpenNMS has grown so much lately that there are a number of things I could stand to learn about the application. Instead, I got to sit on a plane for 7 hours. Envy me.

The HTTP Collector Meets the Weather Geek

Posted on by tarus

One of my favorite new features in OpenNMS is the HTTP data collector. In addition to SNMP, JMX and NSClient, we now can mine data from web pages. In many cases it can be easier to just write to a text file in the web root of a local web server than it is to instrument the data via SNMP or some other method.

I decided to put this to the test.

I am a weather geek. In the summer in North Carolina the usual afternoon thunderstorms can be very spotty, and so while it might rain like mad at the office, the farm (10 miles away) could remain dry, and vice versa. This summer we’ve been experiencing an extreme drought, so tracking rain is even more important. I thought it would be cool to see if it was raining at home from the office.

For several years I’ve had a Davis Vantage Pro weather station. With the associated Weatherlink software it is possible to publish the current weather data on the web, but it wasn’t until I finally got broadband a few months ago that it was feasible to actually use it. This weekend I got around to setting it up and sending the data to a web site every five minutes. The output looks like this:

Now all I had to do was set up OpenNMS to grab it using the HttpCollector. Rather than post all of the steps here, I added it as an example on the wiki.

Now I have nifty reports like:

Note that this won’t work for non-text-based site, like those that use Flash, but for everyone else it should provide an easy way to integrate custom data sources into OpenNMS.

Stack o' Macs

Posted on by tarus

We’re getting ready for the first OpenNMS training to be held at OpenNMS HQ. It’ll be next week and consist of both the three day basics course and the two day advanced course.

In preparation we are setting up our state-of-the-art training facility, starting with a bunch of new iMacs. Ben was tasked with getting Boot Camp installed so we can boot OS X, Windows and Ubuntu on ’em.

Everything looks good and wireless is working. Unfortunately, the ATI Radeon 2400 XP cards in the iMacs don’t have Linux drivers yet – supposed to be out on Wednesday. Let’s hope.

Happy Anniversary

Posted on by tarus

Today marks the third anniversary of The OpenNMS Group. Since we started with no venture money and pretty much nothing but a strong will to make OpenNMS the main management platform of choice, this is a nice milestone.

I have told the “OpenNMS” story a number of times, but never on this blog, so perhaps now would be a good time to tell it again.

I didn’t start OpenNMS, nor am I responsible for much of its beauty. Brighter people than me came up with the original architecture and design, and even brighter people than that keep it growing today.

Since I wasn’t there, my knowledge of how OpenNMS got started is pieced together from what various people have told me, so let me apologize in advance if I get things wrong or overlook someone that I should have mentioned.

OpenNMS was started in 1999 by Steve Giles, Luke Rindfuss and Brian Weaver. Steve was one of the founders of Onion Peel Software, which had just been sold to a company called Netplex. Onion Peel provided software and services around the HP OpenView suite of products, and a map management tool called Amerigo was probably their most famous product. Luke and Brian had both worked with Steve at Onion Peel.

Steve was eager to start another company (he’s one of of them “serial entrepreneurs“) and wanted to stay in the network management space. Now at this time companies like Riversoft had raised £83 million in investment, so it was almost impossible to compete in that space without a lot of money. Steve and company saw open source as an opportunity to develop the next-generation of management software without having to raise tons of capital.

They named the company PlatformWorks and started hiring. Ben Reed, one of the current OpenNMS employees, was employee number 5.

PlatformWorks was acquired by Atipa. Atipa was trying to follow in the footsteps of VA Linux, which had a record IPO. Both companies were trying to become the Dell or Compaq of Linux. This was before it was realized that software drivers aren’t really the issue, and that it’s more about the supply chain and getting that disk drive from Malaysia on time.

Anyway, with the market crash of 2000, Atipa needed a new business plan. They had a warehouse full of 1U servers, and this network management software, so why not create a network management appliance for small to medium sized businesses? Thus Oculan, “Your Eye on the Network”, was born.


The original Oculan OpenNMS team

Oculan used OpenNMS as the basis for its management application, but added quite a number of additional features and interfaces to make it easier to use. They still worked on OpenNMS as a separate project, as even though the bubble had burst people still needed to manage their networks and there was a lot of interest in an open source project for this.

I entered the picture in 2001. I joined Oculan on September 10th in order to build a services and support business around OpenNMS. The next day the world changed, and it became a lot harder to sell anything, much less this new idea of open source network management.

OpenNMS 0.8 had just been released. In the following months I worked to figure out this whole services business and to get some commercial customers. In 2002 we released OpenNMS 1.0, and I managed to sign up our first four customers (two of which are still customers).

Then in May of 2002 Oculan got new investors who wanted to make two major changes. They wanted to focus solely on the appliance and they didn’t want to open source any software. As copyright holders they are free to publish the code under any license they choose, although once something is published under a license such as the GPL it isn’t possible to “unpublish” it.

I believe I could have stayed at Oculan, but in the months I’d been working on OpenNMS I started to really see its potential. To be honest, when I first took the job I thought, hey, HP OpenView without the cost. But OpenNMS is much, much more than that, and without someone pretty much dedicated to working on it, it would die.

So I made one of the toughest decisions of my life. I went down to Steve’s office and asked for OpenNMS. Oculan would keep the copyright to 1.0, of course, but I would become the project’s maintainer and get the domain names, etc. We worked out the details and suddenly I was much more wedded to open source than I ever thought I would be.

That was the easy part. The hard part was telling my wife.

We live on a horse farm. The previous spring her company, Union Carbide, was purchased by Dow, and rather than move to Michigan she decided to leave, spend the summer working with the horses, and get a job in the fall. After 9/11 it was much harder to find a job, so she was working in a tack shop when I decided to do this. I had literally bet the farm on OpenNMS.

The really crazy part was that I didn’t know Java – at all. Sure, I had programmed before, but not seriously for several years. The guys at Oculan were forbidden to work on OpenNMS, since at that time the code base was so similar that there were IP issues, so I was on my own.

Or so I thought.

I started Sortova Consulting Company (pronounced Sore-toe-va even though it meant Sort Of A), bought The Big Black Book o’ Java, and got to work. Slowly I was able to fix some of the more grievous bugs, and slower still a community starting to gel around the project. I got more customers and was able to make my mortgage payment. The community kept me going when I really had no one else to help me, and even today I am humbled by their involvement.

On the farm I didn’t have access to any real form of broadband Internet access. I used DirecPC satellite, which was iffy in the best of times. Getting something like a T1 to the house was prohibitively expensive, and it was actually cheaper to rent an office downtown with DSL. So that’s what I did.

I decided to dump the satellite to save money (my “spend less than you earn” philosophy had gotten me this far) so I needed a dial-up provider. As one of those folks who buys local when I can, I called up the local ISP, Blast Internet Services. Always looking for new customers, I asked the guy who answered the phone what they used to monitor their network. He didn’t know, and suggested I “Talk to Lyle”.

Lyle Estill is one of the most interesting people I know. He’s one of the few people who can make a decision faster than me. To make a long story short, he liked the idea of open source, and since Blast was services company and Sortova was a services company, why not have two struggling companies struggle together?

So I put Sortova to bed and began working at Blast. It was great to have other people working with me. Business grew and OpenNMS kept getting better. In early 2004 we were able to hire two more people to work on OpenNMS, David Hustace and Matt Brozowski.

David and I have worked together at four different companies. There is no other person I’d rather have by my side in business. Even though he had to take a large pay cut and a great risk to come work on OpenNMS, he did it anyway.

Matt Brozowski has forgotten more about Java than I’ll ever know. He worked at IBM when they licensed Java from Sun, and he also did a lot of programming on NetView, so he understands network management. Matt has five kids, so the decision to come work on OpenNMS was also not an easy one, but he also did it anyway.

The three of us now formed the core team around OpenNMS, but a number of other people in the community had stepped up too. The lists now had a life of their own, and I didn’t feel the need to answer every single question.

In 2004 Lyle left Blast to work on sustainable energy in the form of biodiesel. I am honored to know that his exposure to open source communities has helped shape his own. With Lyle gone the three of us approached Blast to acquire the OpenNMS assets, and on September 1st, 2004, The OpenNMS Group was born.

Shortly after that we started the Order of the Green Polo to give an official form to the community, and together we set off to do great things.

So on this anniversary let me say thanks to everyone who made this possible. You helped write this story, now let’s go make history.