DEVOPS WEEKLY ISSUE #112 - 24th February 2013
Not the longest newsletter this week but lots of really good, well written, detailed posts that deserve reading. It’s nearly March, which means it’s just about time for QCon and Devopsdays in London and Monitorama in Boston, all of which I’m going to make it along to. Any readers going along to any of those events let me know.
Sponsor
Devops Weekly is sponsored by Brightbox Cloud - launch cloud servers in multiple UK datacentres in seconds…
http://brightbox.com
News
Fantastic post about culture. This isn’t specific to devops or in my experience technology companies based in San Francisco or anywhere else. Read this, think about where you work and consider what you could do to make things better.
http://blog.prettylittlestatemachine.com/blog/2013/02/20/what-your-culture-really-says/
If you’re using Ruby you’re hopefully familiar with the recent spate of security vulnerabilities in various gems. This appears to have sparked lots of useful tooling and conversations, including this post which describes a way of checking your Gemfile from bundler for gems with known issues.
http://psionides.eu/2013/02/18/pick-holes-in-your-gemfiles/
A recent post about how Heroku deals with load balancing requests kicked up lots of discussion. This post explains exactly how you can predict and design systems for traffic routing, and it does so all in code. Timeline is a network simulator written in clojure and is well worth experimenting with.
http://aphyr.com/posts/277-timelike-a-network-simulator
https://github.com/aphyr/timelike
In what is looking like a great series of articles, this second post about devops in a traditional enterprise organisation does into details about how a software group might be structured. It identifies different teams, and points out some of the problems that most similar organisations face. The next posts promise to look at solutions. Worth keeping an eye out for.
http://niek.bartholomeus.be/2013/02/20/a-closer-look-at-introducing-devops-in-a-traditional-enterprise/
ZooKeeper has lots of interesting use-cases around application deployment, orchestration and infrastructure management. But it also has a reputation as being large and complex. This post looks like a great introduction to installing a single node and using the client to store and retrieve data.
http://blog.cloudera.com/blog/2013/02/how-to-use-apache-zookeeper-to-build-distributed-apps-and-why/
An opinionated take on 12 devops anti-patterns. Everything from the devops team to replacing operations staff with developers to the relationship with agile.
http://blog.devopsguys.com/2013/02/20/twelve-devops-anti-patterns/
Tools
Opsworks is the latest addition to the Amazon Web Services suite. It provides a nicely integrated provisioning environment based on Chef and the EC2 API. If you find elasticbeanstalk not flexible enough, but feel like you’re reinventing the wheel using the low level EC2 API then this might be the perfect fit.
http://aws.amazon.com/opsworks/
zkwatcher, and the related puppet_zkwatcher, are a nice looking solution to storing the list of available services and hosts in Zookeeper, and then using that information in Puppet. This avoids hard coding host information in your configuration management code. The supporting blog post explains how and gives a more detailed walk through.
http://engblog.nextdoor.com/post/34583564058/zookeepers-puppet-masters-and-clouds
https://github.com/Nextdoor/zkwatcher
https://github.com/Nextdoor/puppet_zkwatcher