Month: July 2008

Dev-Jam 2008 – Day 4

Posted on by tarus

Things are really rolling now. We’re heading toward the 10,000th subversion commit since we migrated from CVS last year. That’s a lot of code changes.

I spent the day working on 1.6.0 bugs, and things look on track for a 1.5.94 release toward the end of next week.

For dinner, our perennial Dev-Jam sponsor Papa Johns Pizza stepped up with some free pie. I’ve written about Papa Johns before – they have sold more than US$1 billion in pizza over the internet and for most of that time have used OpenNMS to manage their network.

Plus, I think they make great pizza. Before you flame me, I’m not trying to compare their product with the handmade, oven fired pizza I had in Naples or your favorite pie from the neighborhood pizzeria, but for a large company their product is pretty darn tasty and consistent. They sent one of their Spinach Alfredo Chicken Tomato pizzas today and it is very nice.

Hats of to Chris in Louisville for getting it approved and Ray in Atlanta for arranging everything.

While I’m thanking sponsors, let me send a shout out to Chris and Leslie over at Google (second year in a row they’ve been a sponsor) and to Adam at NetRegistry in Australia who also helped make Dev-Jam happen this year.

Stupid HTTP Collector Tricks

Posted on by tarus

I’m not sure why I like the HTTP Collector in OpenNMS so much. Probably it is because I’ve been struggling for 20 years to get custom data via SNMP and it is just so darn easy to write it to a text file, stick the text file the in webroot of the system’s web server, and then use HTTP to access it.

HTTP is TCP-based, usually on port 80, and most firewalls will allow it, so it is often easier than dealing with SNMP. Plus since there is so much data out there on web sites it can be fun just to grab it and graph it.

I blogged about this once before with my weather station.

Recently DJ asked me how many people were on the various OpenNMS Mailing Lists. I sent him to the Mailing List Summary page on Sourceforge and I mentioned that it looked like our numbers had gone down a little. He said “wouldn’t it be cool if we could track it” and the HTTP Collector immediately came to mind.

If you ever decide that you would like to monitor Sourceforge, be forewarned that they really don’t like port scans or high levels of requests, so I was sure to configure OpenNMS to only gather the data infrequently. This is not a problem since in the week I’ve had the collector running they only updated the values once.

So here is a nice collection of “stupid HTTP collector” tricks which should show off the various things one can do with it. Click on the link to visit the wiki page with the details.

The first is the collection of Sourceforge Mailing List Stats, which results in a graph like this:

I just realized that it might be cooler to store this as a counter instead of a gauge and just graph the deltas, but I’m going to leave it as is for now.

The second is a graph of the major US Stock Index Values over the last couple of months. Sad, I know.



Remember that you can always set thresholds, etc. for data collected this way, although I don’t think I’d base my future day trading aspirations on a system where the data is naturally latent (the NASDAQ.com website doesn’t update in real time).

Finally, here is an eBay Auction Price example. This is taken directly from our Basic Training Class (so those of you coming in September know where to go to cheat).

While these were created pretty much just for fun, the HTTP collector is a pretty useful and serious feature and it is just another way that OpenNMS can easily integrate data from diverse sources.

Dev-Jam 2008 – Day 3

Posted on by tarus

Dev-Jam is more than half over, and things seems to be going well. At least we’ve fixed the issues with Rock Band synchronization by moving the Wii to the projector from the plasma television (which apparently isn’t fast enough to keep up).

(grin)

I spent the day working support and doing triage on the outstanding 1.6.0 bugs. We have around 80 issues to address, but many of them are simply contributions from the community that need to be tested and committed.

We did get the unpleasant surprise of a Sourceforge svn maintenance last night which put us out of commission in the evening. Everything seems to be up and running now.

Dev-Jam 2008 – Day 2

Posted on by tarus

The second day of Dev-Jam went smoothly enough. Everyone has settled in and gotten their development environments sorted. I’ve been running around in my role of cruise director as well as trying to keep our customers happy this week, but I managed to get a little bit more documentation done on the wiki.

In the afternoon Coté spent a couple of hours on Skype with a few people who came out for the conference. I’ll post a link when he gets the podcast up on his site. I was hoping he’d get to talk with everyone, but he was asking enough great questions that we just ran out of time.

It turns out that Monday was DJ’s 30th birthday, so we all went out to a churrascaria up near Marietta. I just think it is so cool that he spent such a landmark birthday with us instead of people he cares about (wide grin).

Dev-Jam 2008 – Day 1

Posted on by tarus

Dev-Jam 2008 is officially underway. Monday morning I gave my “State of OpenNMS” speech, and then turned it over to Matt.

Since last year’s conference we’ve had ten releases of OpenNMS. This was mainly due to Ben’s repackaging of the application into the distro-specific bits (that rarely change) and the main Java code. Ben is an amazing packager (he is an admin with the fink project) and this took our release cycle from two days to about two hours. The best part is that anyone can build their own packages, since OpenNMS doesn’t require commercial packaging tools like Bitrock.

Speaking of releases, next week we hope to release 1.5.94 which will be the release candidate for 1.6.0. If all goes well, we should have a new stable by the end of August. We are then hoping to start work on 1.8 which will focus on capsd and notifications. The goal is a six month, focused release cycle.

We also took care of some overdue business by awarding two people their green polo shirts. Alejandro Galue joins us from Venezuela, and he has worked quite a bit on our thresholding system. Craig Gallen hails from the UK (although he is Irish) and has been instrumental in our work with NGOSS. He has also attended all four Dev-Jams. While they both were inducted into the OGP last year, I wasn’t able to get their polos made until now (I have to do some special stuff with Lands End to get the proper color green).

I then spent a lot of time chatting about the future direction of the application, as well as updating the team on some of the more private aspects of the project. While The OpenNMS Group provides financial support for OpenNMS, it is the OGP that is responsible for its governance, and even though we are open source that doesn’t mean we do everything publicly (grin). There are some amazing things on tap for the next year.

I then turned it over to Matt and we started setting up the projects to work on during the week. We’re using a format similar to barcamp, and we’re tracking it on the wiki. I’m focusing on writing the new OpenNMS book, which is also available on the wiki.

It looks like it is going to be a great week.

Dev-Jam 2008 – Day 0

Posted on by tarus

Well, technically it is day one, as we are quickly approaching 1am on Monday as I write this. I’m pretty tired, so this one will be short.

We just picked up Bill Ayres from Hartsfield. He brings our total to 16 for the entire week, although we have some people who are supposed to drop in for awhile.

I started off this morning at home. Alex and I took the Magic Bus (a 15 passenger van we’ve rented for the occasion) over to Dave’s where we picked up him and Ben, then up to Matt’s house before getting on the Interstate. Something like 8 short hours later we arrived a Georgia Tech.

After unloading, we proved that 15 people is about the maximum you can fit in a 15 passenger van as we headed out to The Vortex for dinner.

Alex has some pictures from the day up on his blog. In general expect some serious blog activity over the next week. Be sure to check out the Planet OpenNMS for an aggregation of most of them.

Dev-Jam, News, OUCH and echo

Posted on by tarus

Well, things have been a bit crazy around here getting ready for Dev-Jam.

This will be our fourth developer’s conference and it is one of my favorite times of the year. This year we are moving to Georgia Tech (a change from UMN) and I can’t say I’m looking forward to being in Atlanta in high summer, but the rooms we’re using have air conditioning and bandwidth so what more could we want.

We also have people from five countries in attendance. In addition to the USA, we have two guys from the UK, one from Venezuela, one from New Zealand and Alex – who is German, yet works in Switzerland and lives in France – I’ll let you choose which country we should count for him.

Almost all of the OGP was able to make it, so it should be a productive week. Stay tuned for updates from the conference starting this weekend.

Last Thursday I was invited to Michael Coté and John Willis’s weekly podcast. John was a bit late coming to the party, so Michael invited a friend of his, Matt Ray of Zenoss, to join us. I got to talk about OpenNMS (imagine that) as well as my current distaste for the overexposure of the term “cloud computing“.

When John joined the call he tossed OpenNMS a bone by bringing up software licensing. One place where OpenNMS differs greatly from other “commercial open source” companies is that while support and services are available, the software is 100% free and open. Matt’s company publishes some of its code under the GPL (which Matt quickly pointed out) but all of their “non-community” code has a proprietary license.

Matt seems like a nice guy, so I didn’t rise to the bait, plus he had a cold. Most “commercial open source” companies have a business plan that relies heavily (if not entirely) on software licensing revenues. Since it is difficult to sell open source software more than once, software developed for this model can’t be free and open. Since a smart business wants to maximize profit, these business should be finding ways to drive people from the “open” solution to the proprietary one. Most importantly, a smart business will make decisions concerning their “open” solutions to maximize the migration to proprietary software.

This isn’t a bad model, or even a wrong one. I just grouse at using “open source” in any fashion to describe it.

Matt did give us a compliment by mentioning that the new Zenoss Masters program is modeled on the OGP. Building community is fun and we wish them the best of luck with that.

The biggest comments so far on the podcast have been about the ability of OpenNMS to replace portions of Micromuse (now IBM) Netcool. I’ve always been a big fan of Netcool, and I worked with Micromuse even before they had an office in the US. But the price was always astronomical, so a couple of years ago we decided to implement some of the functionality of Netcool into OpenNMS. We have a number of clients who have replaced Netcool, but we are currently working with a large telco in Italy that will be our biggest challenge to see if we can take a very large Netcool install and replace it. David is spending several weeks a month in Italy with Antonio for the rest of the year to implement some of the migration. It’s hard work, but fun.

Both John Willis and Doug McClure posted about it on their blogs, so I hope to be able to send them some actual numbers in the near future.

Hrm, lessee, what else is going on …

OSCON is this week. I went last year, but with Dev-Jam next week I just couldn’t justify going this year. Plus, it is really more of a coders show and I’m not a coder.

LinuxWorld is the week after Dev-Jam. They have been kind enough to give us a booth for a couple of years, but the show has become too much “commercial open source” for my tastes, so we didn’t ask for one this year. Jeff is going there to work in the Open Solutions Alliance (OSA) booth. While we’re not a member of the OSA, OpenNMS integrates with both Hyperic and Concursive, which are members, so we’ve been asked to demonstrate this interoperability. David decided to call this integration the “OUCH” stack (grin). It stands for OpenNMS, Ubuntu, Concursive and Hyperic. If you’re going to the show be sure to say “hi” to Jeff.

Finally, I want to send a “shout out” to Eric Bradford at echo. He sent us a card this week for the Fabulous, Amazing, and Incredible Wall of Cards. I’ve included it below, as it made my day.

OpenNMS: International App of Mystery

Posted on by tarus

I am constantly amazed at number of people outside of the US who use OpenNMS. According to Google Analytics this month we’ve had 45,738 visits from 180 countries/territories.

13,037 of those came from the US, which means that over 71% of our website visitors are from other countries. Now I’m not sure how accurate these metrics are, but we are seeing at the OpenNMS Group a lot of international interest (we have clients in 17 countries now).

Dave is currently in Italy where we are working with one of the largest telecom providers there to improve their network management capabilities. Carriers spend a tremendous amount of money on solutions that usually take years to deploy, yet we should be able to deliver a number of large improvements in under six months. With no “per node” pricing of the software, OpenNMS can grow to meet their needs with only a small incremental cost associated with the additional consulting and development. In their highly competitive market this is a good thing, and it demonstrates the value that free and open software can bring to the largest of companies.

As for me, after Dev-Jam I’m off to Australia for a month. We have a number of clients in Oz, but this will be only my second trip there. With e-mail and Skype we are able to handle most support issues without having to resort to heroic effort or odd hours (we get the ticket first thing in the morning our time and it is solved by the time they return for their morning). We’ll be working with a large department of the government of New South Wales.

This week Ben is in the Amazon in Brazil. While Brazil is a hot bed of open source use, we don’t have a client there (yet) – this is just vacation. Hanging in the rainforest – that’s just how we roll.

(grin)

Sticker Shock

Posted on by tarus

It looks like I am going to be in Australia for the month of August.

I love Australia, and when I was there in 2004, Sydney replaced San Francisco as my favorite city. But the 15 hour flight from LAX to SYD is murderous. Seriously, I can deal with long flights but 15 hours in coach is painful. On the way back I watched Return of the King three times in a row and the flight was barely half over.

So I thought, hey, this time I’ll fly in business class. Yeah, it is more expensive, but should be worth it.

I went and looked up the price on American: over US$19000.

What? That is the price of a car. For this post I decided to just price a flight in business from LAX to SYD (versus all the way from RDU). It came to US$18700 with taxes, round trip.

If we assume that the round trip is 30 hours or so, that comes to around $623/hour. That’s outrageous. You can get over three hours of top-notch OpenNMS consulting for that. Heck, Elliot Spitzer doesn’t spend that hourly rate (well, at least for 15 hours straight).

Who pays this? I really can’t imagine doing it. Well, I mean if I was Tiger Woods and made millions, US$19K might not seem that important (and I’d probably upgrade to First for just $5K more), but from a business standpoint?

I had a friend say “well, just charge it to the client”. I did think about it, and if it was under $10K I would have even suggested it, but it was almost twice that. For the difference I figured I could fly out in coach a couple of days early, stay in the most expensive hotel room in Sydney, dine on endangered species and still come out ahead.

Sheesh. $19K for a plane flight (well, two). So I’m in coach for a fraction of the price, heading out a couple of days early (staying at the Courtyard) and the most exotic thing I’ll eat is kangaroo, but at least the client will come out ahead, which means they can buy more OpenNMS stuff.

So, am I an idiot for not charging the client? What’s the most expensive plane ticket you’ve purchased?