Monthly Archives: July 2011

DevOps Downunder Melbourne 2011 day 2

The second day of DevOps Downunder drove home for me at least the message that DevOps is just as applicable in small 4 person startup environments with a couple of servers as it is in large companies with complex heterogeneous environments.

Robert Postill spoke about his company’s challenges developing an appliance that talks to in-house data warehouse apps.

With this kind of setup Continuous Deployment and Monitoring for example are more difficult. They are happy New Relic customers. (As were many speakers today!)

Leni Mayo presented a talk on a 6 month old baby startup

When there are no Ops on staff, Devs do the Ops. Like: going on-call, provisioning servers, monitoring, deployments (old school Ops I’m talking about).

Configuration Management is close to the heart of progressive Ops types. Puppet is a popular tool for CM these days. Garrett Honeycutt from Puppet Labs presented some background on Puppet and some tips for using Config Management effectively.

Next Wade Millican from Ninefold talked about Cloud Bursting to prevent outages. But he also went over a lot of great advice from the sysadmin perspective about building resilient, scalable, maintainable, secure infrastructure.

Props to Ninefold for the best Schwag for the conference 🙂 Beanies and skivvies were appropriate for cold cold Melbourne. (Sorry Mr Google. Your thong schwag was lame as.)

Open Spaces after lunch included a really detailed look at DevOps at from Colin Panisset Sounds like they’re “doing it right” on a complex site with lots of legacy components. In this kind of environment DevOps includes rotating sysadmins in and out of development teams AND rotating developers through Ops teams. This kind of idea cross pollination has enabled much closer collaboration.

Also Patrick Debois and (assistant whose name escapes me, sorry) demo’d Vagrant, VeeWee and McCloud.

If you’re not using Vagrant go get it. Now. Go on. It’s cool. Vagrant drives Oracle VirtualBox and enables the creation of consistent virtual development environments.

The final and most heated discussion of the day was entitled “Why is recruiting good people so hard. Conversely why is finding a good job so hard.”

I have lots to say on this topic. And in this case choose not to.

Conference was closed by the esteemed organizing gentlemen Chris Bushell and Lindsay Holmwood.

Awesome work guys.


Leave a comment

Filed under Uncategorized

DevOps Downunder Melbourne 2011 day 1

It was awesome to have a DevOpsDays conference right here in my home town of Melbourne. I was super excited to be going. The 2 day conference was sold out. This alone shows the amount of interest in DevOps at the moment.

The views below are my own and do not represent the opinions of my employer.

Day 1 kicked off with a welcome from organizer Lindsay Holmwood.

The keynote was by none other than the “Father” of DevOps Patrick Debois.

His slides are here What’s in a word

A few people are getting too hung up on what the word means. Is it too exclusive? What about the Business, Test/QA, Network, Storage etc etc… people. Is it a job title now?

I really like the term “Frictionless IT” that sums up what DevOps is.

Does DevOps need a manifesto? Patrick doesn’t think so. Try reading the Agile manifesto with an open mind and think holistically, rather than focusing on software. i.e. Working infrastructure is the primary measure of progress.

Can DevOps work within an ITIL organization? Yes it can. If you don’t let bureaucracy get in the way too much.

Second up was “Kanban for IT Operations” from Jason Yip from Thoughtworks.

His slides are here Lean and Kanban for IT Operations

A great overview of Kanban and why it works for Ops as well as Dev. I heartily second this.

There was an interesting question from the audience “what is a good rule of thumb for Work In Progress (WIP) limits”, the logical answer is “measure the effect of changing the WIP limit and see what happens”.

3 tasks per person was suggested. I’d start with 2. The objective is to keep the number of tasks “in progress” small, and throughput high. Don’t be afraid to move tasks out of the “in progress” state.

Jason’s message for me was don’t be afraid to experiment with processes. Measure the effect of changes and optimize.

Next was Nish Mahanty from MYOB talking about “Building High Performing Teams”.

Takeaways for me were the importance of aligning whatever motivates dev and ops. Also sounds like MYOB know the importance of connecting with their customers. Nish said ops/dev guys listened in on support calls from users of their products to get a better feel for their pain points. Awesome stuff.

Next was a panel discussion. Link to video coming?
Unfortunately there are no silver bullets. Or silver guns.

After lunch were the Open Spaces sessions.

I’m sorry, but I don’t find these are as satisfying a format. I really like 5 minute Ignite talks more.

It was interesting to hear Google’s take on DevOps via Paul Cowan. The Site Reliability Engineer (SRE) at Google is really the glue that keeps everything running. Google also have Software Engineers and guys who deal with the hardware… more traditional System Administrators, Network Administrators etc…

But the SRE guys are “practically perfect” 🙂 They know about how Google works from a holistic perspective that the developers mightn’t have. They mentor developers. They cut code. They go on-call. (This sounds terrifying.)

Google puts developers on-call for the first weeks of launching a new product. Brilliant.


Filed under Uncategorized