3 minute read

From chatops to unikernels and from getting started with application deployment to failure and human factors. Hopefully something for everyone this issue.


Devops Weekly is sponsored by Brightbox Cloud: the multi-zone cloud platform built for High Availability.

Try an SSD cloud server in 30 seconds with your £20 free credit…


7 Practices to Expand Performance and Effective Collaboration in DevOps

In DevOps, the product is only as good as the teamwork behind it. In this webcast Mark Tomlinson, performance engineering veteran and founder of the popular PerfBytes podcast, shares 7 practices to help you expand performance and effective collaboration into your DevOps team.


The concept of chatops has been around for a while, but implementing it can involve building, maintaining and integrating a lot of code. Enter cog, which though early features some nifty ideas like command pipelining.

Docker as a universal packaging format is interesting, but you still need to be able to call the software in the container from other software. This post shows one way of going just that from Python.

A good exploration of the pros and cons of the single address space as seen in unikernel systems. A good primer on why this is a design choice with some advantages, rather than something that’s a side effect or an afterthought.

A detailed side-by-side comparison of Chef and Ansible. Covers basic differences and differences in language, along with lots of examples.

A nice deck making it clear that it’s completely possible to use continuous delivery practices in a WIndows environment. Includes several examples as well as recommendations.

A great talk video focused on introducing more automated deployment workflows to web developers currently still using FTP to upload files.

Failure, and in particular how human/computer systems fail, is a recurring theme in devops-minded conversations. This post explores so-called zero harm and never events, and tries to explain why avoidance of accidents can’t be the only strategy.

A thoughtful post making the claim that distribution packaging as it stands is a broken model. Focuses on the potential for the gap between upstream developer and packager to lead to missed security updates or diverging, hard to manage, codebases.

A useful review of the paper IncludeOS: A minimal, resource efficient unikernel for cloud systems. IncludeOS is a new C++ based unikernel base with some interesting ideas.

A second post in this series on chaos engineering, focusing on the shift in mindset required and the non-technical parts of selling the idea of introducing failure to your team and organisation.

A nice set of examples of separating code from data in the context of infrastructure automation. In this case with examples from Salt, Ansible and Puppet.

A technical post looking into the details of how to run an elasticsearch cluster at scale, in this case ingesting around 25k events per second.

A quick ode to log files, and in what situations just looking at the logs can be the simplest solution to finding a particular problem, over more complex tracing tools.


Devopsdays Austin is back with the next event on the 2nd and 3rd of May. There are a handful of tickets left and the CFP is open until the 25th of March.


DbDat is a tool for checking the security of your database, in particular working out which CVEs the database in question may be affected by. It has support for scanning everything from Postgres to Oracle and DB2 to Mongo.

Devops Weekly is sponsored by Brightbox Cloud: the multi-zone cloud platform built for High Availability.

Try an SSD cloud server in 30 seconds with your £20 free credit…