DevOps Days Galway
A look back at the successful conference
Outside of work, I normally blog about music related subjects here on Medium, but I just recently published some thoughts from my profession in operations support, on DevOps which is a subject I feel passionate about.
Subsequently, it was great to see that the first Devops Days event in Ireland kicked off in Galway and I went along to find out what the latest thinking and experiences are in this area.
HPE’s Darragh Bailey opened the conference with a friendly welcome note, laying out the agenda and highlighting what to look forward to.
Over the two days, a wide range of experienced speakers addressed the community on a host of DevOps related topics. There was some stand out presentations for different reasons, but the talks that resonated most with me were the ones that came from an open engineering ethos.
Examples included the following:
One engineer model
Mary Newell from Cisco went into good details in her “Dragging the dinosaur into the 21st century” session which explained the journey for her team and the drivers for it. It was pretty evident in the timeline she presented the challenges of adopting DevOps. The model that works for her team after their journey, is an ‘one engineer’ with no dedicated operational people, but each developer takes a operations focused week in rotation. Not only that, but Dev and QA roles were merged, which drove a mind-shift change. This was not without some challenges, shared code base, test coverage and stability, not being in control of the entire environment, troubleshooting skills and her advice for this approach included:
- Pair programming with Dev and QA
- Control your environment (as much as possible)
- Test stability
- Invest in metrics
Shane Touhy, also from Cisco, was another person to highlight the need and the benefit of investing in application metrics. Shane shared details of the microservice they developed, to periodically query events from different sources and combine them into more useful, friendly metrics to help understand their customers’ experience, particularly critical when supporting a cloud based voice and video platform. Shane demoed their Kibana usage, this was something that Ross Coll from DXC was also putting a case for, with their demo of how they use the Elastic stack for a single pane view of their complete delivery life cycle.
End to End Testing?
Peter Kennedy from Pager Duty, delivered a highly entertaining and informative session called ‘Silent, But Deadly — Production End To End Testing’ where he explained in an open engineering talk, the challenges they faced with complex failures and the strategies that PagerDuty used to always make sure their service was always up and running. His opinion on end to end testing was interesting, he explained how it makes sense in test but not in Production and he also pointed to a Google paper by Mike Wacker .
Peter explained the pain points of monitoring distributed systems and his take away advice was to derive system health from leveraging metrics and without writing tons of code!
Poke it until it works in production
Another excellent session came from Demonware’s Tom Shaw, who delved into his experiences with building a containerised continuous delivery pipeline using Docker in his Skypilot project and he shared some tips for container adoption including building a coalition of users, learning from growing pains and the change in thinking;
“The most important thing about containers is the process of using them, not the things themselves” Justin Warren
With containerized pipelines he discussed taking the fear out of deployments where the emphasis on making them simple/fast/reliable/safe, to where they can be even tested by non technical project managers. He added that they continue working to reduce the entry level to service deployment. Basically, “give the dev team powerful tools and let them get on with it”!
Serverless compute platform for stateless, event driven code
In his ‘Thunderstruck by Serverless’ talk, Conall Bennett, provided an insightful view to how his team fell into a serverless implementation without any real experience of the technology and how they learned about it along the way. Interestingly, they use exploratory performance tests to determine and quantify language deficiencies on the platform, in order to help drive our implementation approach. He made some stellar and pertinent points, about changing the focus to delivering code and not worrying any longer about operations issues
- Constantly changing landscapes and software architecture can cause fatigue, even for the tech curious.
- In reality, “Lift and shift”, sometimes mean “re-Architect” (And “re- Architect” again).
- Fully integrate the performance tests into an automated performance pipelines
Reliability at scale
This talk was about what problems GO-JEK India (GO-JEK is South Asia’s fastest growing transport, logistics, hyper local delivery and payments startup) faced and how they handled those issues with proper planning, building tools and SRE culture.
Their lessons learned along the way
- Reliability is an iterative process
- Developers should develop, deploy, and own reliability
- Allow people to make mistakes
Technology topics such as Kubernetes came up in a couple of talks, where HPE’s Cynthia Lopes do Sacramento and Miguel Castilho Dias demoed the advantages and benefits of Hybrid Cloud with Kubernetes Federation and Konstantin Semenov with his Deploying a maintainable Kubernetes cluster in 20 minutes.
Ignite Talks
Ignite Talks are a series of 5-minute presentations, where each presenter uses 20 slides, which auto-advance every 15 seconds.
Tony Chapman encouraged attitude over skill, “People, Process, Technology, 3 of the key things to consider if you’re looking to implement DevOps principles.”
Pivotal’s Garima Sharma spoke of Devops Team always being under resourced and their train/driver/track to alleviate some of that, for multiple release deployments.
During these talks, many novice presenters got up on stage, for their first time. Mark Walsh, for example, made his debut with an introduction to MetLife’s current technical landscape and the road map for their Global Technology Campus in Galway, where they have been adopting DevOps principles and practices from Continuous Integration to Continuous Delivery and Containerization to Orchestration.
Open Spaces
The conference also had a open space, where topics were raised for discussions and were voted for.
People shuffled sessions around, merged sessions as attendees decided with their feet, on which space to go.
These open discussions turned out to be friendly, informative and engaging. I found it validated thinking in part on subjects and raised questions in other topics. It was good to hear practices, technologies and common challenges that are being faced in the industry, in other companies.
Overall, the event went very smoothly for the attendees, much in thanks and credit to the organisers including:
Darragh Bailey, Xabier Pedrera, Jon-Paul Sullivan, John Fitzpatrick,Emil Dimitrov, Graham Hayes, Emma Meehan and Waldemar Znoinski
