DEVOPS WEEKLY ISSUE #187 - 3rd August 2014

Today’s issue is a little later than usual as I’ve been on holiday for a few days. Lots to catch up on, with lots of detailed posts about peoples individual workflows this week.

Sponsor

Devops Weekly is sponsored by Brightbox Cloud - serious UK-based cloud infrastructure from only 1.5p per hour (£10.95/month)

Start your £20 free trial now: http://brightbox.com/devopsweekly

News

A good delve into scaling graphite, in this case looking at minimizing the lag for certain metrics being queryable with Kernel-level I/O tuning and other tricks.
http://grey-boundary.com/minimizing-datapoint-lag-in-graphite/

A great intro to various aspects of systems thinking with regards complex web based systems, with tips for embracing complexity, designing for resilience and testing in production.
http://www.slideshare.net/jurquhart/gluecon-keynote

A good post on the issue of security and linux containers (and docker in particular). Clearly explains the current differences between containers and virtualisation based on kernel subsystems and namespaces.
http://opensource.com/business/14/7/docker-security-selinux

A nice walkthrough of a particular infrastructure workflow based on Packer, Chef and CloudFormation. Lots of reasoned detail around moving from development to production.
https://medium.com/@pereiragoncalo/continuos-integration-and-delivery-with-packer-chef-aws-d29453d93382

If you’re using AWS you’ll probably have made decisions about which regions and zones to use for different workloads. This post delves into some simple benchmarks for upload speeds with some interesting findings.
http://www.takipiblog.com/2014/07/30/aws-speed-test-what-are-the-fastest-ec2-and-s3-regions/

A nice piece pondering on the relationship between infrastructure acceptance testing and monitoring. I still think there are lots of good patterns to discover here.
https://devblog.timgroup.com/2014/07/30/mdi-monitoring-driven-infrastructure/

A good reminder, with lots of code examples, that Vagrantfiles are simply Ruby. A few tips for tidying up multi-node configuration and using chef-zero with Vagrant.
http://christinemdraper.wordpress.com/2014/07/23/multi-node-topologies-using-vagrant-and-chef-zero/

A very detailed step-by-step guide to setting up a continuous integration pipeline with Jenkins. Touches on the intended workflow for people using the pipeline as well.
http://devops.judes.co.il/post/92622747539/continuous-integration-pipeline-an-example

Jobs

We thought the process of finding a new DevOps job needed to be shaken up. So we built Hired, the first true marketplace for engineers. Join a Hired one-week auction and let 1,000+ companies compete for you.
http://hired.com/?utm_source=newsletters&utm_medium=devopsweekly&utm_campaign=aug3_14-devopsweekly

Evernote is hiring! Our goal is to be an ever-innovating, globally-inspired, benefit-obsessed company for the next 100 years. This requires a singular focus on making the absolute best and smartest product possible. That means, we need talented individuals, like yourself, to help us.

Build and Release Engineer: http://evernote.com/careers/job/?id=18526&gh_jid=18526 Senior Operations Engineer: http://evernote.com/careers/job/?id=15532&gh_jid=15532

Events

DevopsDays Toronto is coming up on the 12th and 13th of September. The call for proposals has just closed so expect the programme shortly. Still room for sponsors too.
http://devopsdays.org/events/2014-toronto/

Another DevopsDays event, this one in Chicago. The call for proposals is still open until August 15th with the event scheduled for October 7th and 8th.
http://devopsdays.org/events/2014-chicago/propose/

Tools

Managing local machines with Vagrantfiles has lots of advantages, but when recreating that environment in production you might turn to something like libcloud. Duplicate no more, with with vagrant driver for libcloud.
https://github.com/carletes/libcloud-vagrant

Terraform is a tool aimed at giving you a single view of your entire infrastructure in code. Describe instances and supporting services (like DNS) via a domain specific language and use the command line tool to launch across multiple infrastructure providers.
http://www.terraform.io/

Consul is an orchestration mechanism with fault-tolerance based on the gossip protocol and a key/value store that is eventually consistent. Hiera-consul will allow hiera to write to the k/v store for metadata centralisation and harmonisation.
https://github.com/lynxman/hiera-consul