4 minute read

Monitorama in Boston was amazing. Huge thanks to Jason Dixon and everyone else involved who organised, attended or sponsored the event. If you’re not interested in monitoring I’d probably recommend skipping this entire issue, if not the links, presentations and new tools should keep you buys. Even the jobs are with folks who were at monitorama. See everyone in Berlin.


Devops Weekly is sponsored by Brightbox Cloud - launch cloud servers in multiple UK datacentres in seconds…


If the huge amount of content from monitorama in this issue makes you wish you had made it along then you might be in luck. Monitorama is coming to Europe, Berlin, later this year. Sign up for more information.

A talk in the style of an essay, this one on Monitoring for humans. A good account of the status quo, lot of improvements to the technology and tools, but not as much discussion of what makes a good check or what makes a worthwhile alert.

Slides from one of the workshops held on the second day. These are from the statsd workshop, providing a quick introduction, showing examples, talking a little about writing backends and suggesting a few nice future enhancements.

An entertaining run through the state of monitoring, focusing on a few areas where we could be doing a lot better or thinking about a lot more; for instance security, data retention, alert fatigue, visualisation patterns and more.

More slides from monitorama, these ones from Dr Neil Gunther. Features a great history of monitoring and lots of science and statistics which we should be applying to our tools and approaches.


The videos from Devopsdays London are all up now. Lots of good talks worth watching, in particular if you’re interested or working in a large organisation as that turned out to be something of a theme.

Good list of things to consider when writing configuration management code, aimed at preventing a proliferation of very similar modules and improving prospects for sharing code.

If you’re into podcasts then here’s another one to add to the growing webops/devops shows. We love monitoring is a podcast dedicated to all things monitoring. The organisers are looking for people to go on the show too.

Article positing that we need to see much better ways of testing performance and scalability in a world of continuous deployment. The comments are interesting too, mainly highlighting resistance to the idea within large enterprises.

A follow up to a previous article, about problems that occur if you take a pipeline approach to monitoring. Specifically, the case with checks that are based on collected metrics, and the different failure modes.


Help build the platform that powers a better, fairer banking experience at Simple. Join a talented team that chooses its own tools (and open sources new ones); works across web, Android, iOS, and Ruby/Scala/Clojure backend apps; and develops a secure and scalable banking service on AWS.

MIT Lincoln Laboratory is currently looking for a student enrolled in college to do a paid internship/co-op in our Advanced Technologies division. This is a great opportunity to get your feet wet in a role that encourages creative approaches to operations work. You should already be pretty comfortable with Linux environments for this internship, and be familiar with the configuration management platforms such as Puppet. Knowing a language (like PERL, Ruby, Python, PHP) is also useful, as we’ve got a variety of user-facing and internal tools that could use some improvement. Unfortunately, due to the nature of Lincoln’s work, this internship requires US Citizenship.


One of the talks at Monitorama, from Jesse Newland from GitHub, talked quite a bit about their new pagerduty integration with their hubot chat bot, and how it helped with being on call. The code for that plugin is now open source.

Norman is a dashboard for puppet reports, based on information sent to logstash and stored in elasticsearch. If you’re already using Forman, but only for the report dashboard this might be for you.

Another of the monitorama hackday projects was a way to build resilience into statsd using a proxy with a consistent hashing ring. Described as a proof of concept at the moment but we saw this working on the day.

Dashing is a new dashboard framework from Shopify. It’s written in Ruby and Coffeescript, already has impressive looking documentation and a good demo, and looks a great way of quickly building attractive dashboards.

Ever wanted to update your TV screen dashboards using your chat bot? Crimson Twins is used by Mozilla to do just that, allowing you to easily change what’s been shown on your dashboards based on what is relevant to the issues at hand.

Mixing business metrics with operational data is a great way of fostering collaboration and helps with prioritising the right things. Anthracite is a simple dashboard tool for submitting and showing those events, with nice graphite integration so you can mix this data into your existing metrics.