Written by David Stewart, Mike Tomaras
It was 3.45am when my alarm went off on Monday morning. After a few minutes of wondering what was going on I somehow managed to get myself ready and start my journey up to DevOps Edinburgh. The conference was held at Dynamic Earth which Mike described as a “mini O2 arena”, a name I’m sure the Scots would have loved.
The program took place over two days. The mornings were broken down into six half hour talks, followed by small five minute talks called “Ignites”. It was interesting to see how much information each speaker can pack in a five minute time span and also how fast they can actually talk! In the afternoon they had three Open Space sessions where attendees were invited to come up with their own discussion topics. The ones with the most upvotes got a room and 30 minutes for anyone to join in, ask questions and share war stories.
One of the common topics from the conference was around company culture and the need to create a positive working environment where engineers didn’t fear making mistakes. This was covered in the opening keynote from Maria Gutierrez, VP of Engineering at FreeAgent and Gerie Owen, a Software Test Manager at Qualitest Group. Simple notions we can look for and improve include trust, managing conflict, having accountability and then the results come easy. We can also look for warning signs of an ailing culture, one of them was talked about a lot and it’s commonly known as Groupthink.
Jason Hand’s talk about proper post mortems also gave a good insight as to how we can actually get a lot more actionable points from a single outage that are not simply tech related. For example a simple outage can mean that your on call, paging, monitoring and site fallback pages need work as well. It’s not just the product that the end user experiences, it’s your whole response process.
Finally Jenny Ducket showed us how to run a workshop the agile way. You don’t have to be an expert to talk about it. Like with code, start small, get feedback and iterate from a small code kata all the way up to a two day conference!
The other common topic was around containers and Kubernetes. Having never read anything about Kubernetes I did a bit of research around what it actually was. Essentially, Kubernetes and ECS solve the same problem, managing containers across a cluster. As long as we are using AWS to deploy containers, there doesn’t look to be an immediate need to learn Kubernetes while ECS is available. It was interesting to see that Docker and containers were definitely at the forefront of the operations world and that there was no mention of serverless or any improvements on traditional deployment setups.
Another interesting talk was entitled “The Perfect DevOps Storm” by Paul Gillespie who is the Developer Enablement tribe lead at Skyscanner. He described how Skyscanner evolved to cope with releasing software hundreds of times a day. He talked about the importance of logging and getting feedback once a feature went live. The warnings he highlighted with moving to microservices included introducing un-intended technical debt by giving developers free reign on the technology they use for each service, making it harder for operations to handle the whole system and new hires to get up and running fast, something that I could easily see happening. His talk was a balanced view of all the pros and cons of using microservices and showcased the fact that it takes a lot of work and discipline to achieve a good result. There are no silver bullets.
Security got its fair share of coverage throughout the talks, with the talk “Devops state of mind: continuous security with Kubernetes” by Chris Van Tuin. The point being that security, like QA, is better built into the product from the start. We can use the CI process to test against known vulnerabilities and catch security related problems before they ever get out to production. The mere use of containers has several security benefits over traditional servers as well (i.e. they are recycled regularly, so any compromised instances get trashed).
Overall I thought the conference was good and definitely gave me topics I want to research further. The positives included the talks on culture and engineer burnout along with finding out a bit more about containers and Kubernetes. Unfortunately it seemed like the Dev in DevOps was a bit underrepresented in the crowd and the talks. Getting an operations point of view was still interesting; but I felt it could have been more balanced.
The final takeaway I had was around engineer burnout. As the industry moves more towards DevOps, there seems to be ever more pressure on developers to learn deployment related skills (Docker, container orchestration, Terraform, AWS etc). This is something all companies that want to move that way should also take into account, i.e. supporting the learning of the ever increasing and changing technology their employees have to use. I personally enjoy acquiring new skills and am fortunate that Tombola encourage protected time to improve them.
- Continuous Delivery
- SRE by Google
- Patrick M. Lencioni books on team management: here and here
- The Phoenix Project