I am here in lovely Portland Oregon attending Monitorama. Monitorama is a 3 day open source monitoring convention.

Monitorama had catered food, and drink. The food was plentiful and delicious, and the drinks were amazing.

There were 10 talks, I have made a quick summarization below. I don’t have time to write in detail about each one, but I am sure you will get the gist from the basic summary.

Please, no More Minutes, Milliseconds, Monoliths… or Monitoring Tools!

The first talk from Adrian Crockford was about monitoring by the second as opposed to by the minute. Gene Kim’s notes summarize it better than I do, but in short: focus more on monitoring tools for developers, monitor by the second, and learn to analyze the data properly.

Computers are a Sadness, I am the Cure

Jason Mickens provided the funnest talk of the day. I couldn’t fully grasp the message, but In short Bane == NoSql, Cloud == Crazy Train and stop talking about map reduce.

Simple math to get some signal out of your noisy sea of data

Toufic Boubez Provided the third talk of the day. Talking about (not so) simple math around your data. He urges people to stop treating data as Gaussian, and truly understand how their data gets arranged.

He pushed for using algorithms with sliding means as a way to truely understand your data. Just remember you spent time collecting it, you should spend time analyzing it.

The Care and Feeding of Monitoring

Great talk from Katherine Daniels, about how everyone needs to work together better to monitor things. She noted that no one sets out to make bad decisions, but bad things sometimes happen because of decisions made.

We take some things, and put internets on them ~ Dan Slimmon

Dan Slimmon’s talked a lot about alarms. Smoke detectors people pay attention to, but no one pays attention to car alarms.

He talked a lot about how to actually use probability statistics to understand how often someone will get phone called on an actual issue or not.

Metrics 2.0

This was a strange, yet insightful talk about how metrics should be more self decriptive. Dieter Plaetinck wants people to try to use a new standard in metrics. I’m not sure how I feel about metrics 2.0, but I do agree that metric names can be ambiguous.

Our Most Wicked Problem

…A talk about the diversity problems in the industry today. Ashe Dryden talked a lot about how women are far more excluded than included in IT. This was an insightful talk, about how we need to provide space to all of the women coming into IT.

The cost and complexity of reactive monitoring

Chris Baker talks a lot about how reactive monitoring is terrible, and proactive is the way to go. Most humans never run though a logical sequence of troubleshooting steps, but tend to blame the newest thing they put into production.

From Zero To Visibility

Bridget Kromhout finished up the day with a talk about setting up the monitoring in her company from nothing. Going from a bunch of random shell scripts, to a proper monitoring framework (nagios, graphite, statsd, pingdom, etc.). Some of the pain points she encountered were around budgetary restrictions.

After party

The after party was great, the music was quite loud which made it hard to talk to people. I talked quite a bit to the New Relic crew about what they were doing. The one thing I was blown away by was how welcoming everyone was, and how everyone were fighting the same problems in their respective companies.