Operability.io 2016
On September 19–20, a group of us from codebar attended Operability.io, an annual conference about DevOps organised by Marco Abis. This year’s conference took place at one of the most beautiful venues I’ve seen at tech events — the Milton Court Concert Hall at the Barbican. This is a guest blog post co-written by myself and Vanessa Virgitti.
As someone who is relatively new to Ops but has been Dev-ing for about two years now, I was really impressed with the breadth of topics covered as well as the charisma of all of the speakers. We loved hearing all of the talks. Here is a small selection of what we learnt:
Mark Burgess (@markburgess_osl) kicked off the day with an academic discussion about assessing outcomes in distributed systems. Promise Theory offered a set of principles for reasoning about the responsibilities and capabilities of discrete components in a system. When viewed through the lens of Promise Theory, many microservice-based architectural designs can be shown to have wrong component boundaries.
Sarah Wells (@sarahjwells), the Lead Engineer at the Financial Times, told a story about the FT’s journey from siloed, piecemeal deployment tools and processes to one that was much more collaborative and intuitive, reflecting the DevOps philosophy. Spoiler: It wasn’t easy, and it didn’t happen automatically. At the FT, this transformation required deliberate effort towards integrating Ops engineers into every product team, getting buy-in from all the teams, and encouraging all teams to adopt a “You build it, you run it” principle. Some obstacles they faced included managing out-of-hours support for small teams and projects written in languages not widely-used throughout the department. It was refreshing to learn that even a large, successful company like the Financial Times faced many setbacks on their road to creating a healthy DevOps culture.
Casey West (@caseywest) from Pivotal talked about cloud-native operability, which spans an entire ecosystem of other topics including microservices, continuous deployment, and DevOps culture. The slide that resonated the most with us was this litmus test of what constitutes a genuine microservice:
A microservice is an application small enough that an engineer new to the source code can reason about it in a day or less.
His talk was chock-full of humorous (and true) insights about continuous delivery and platform automation. An automated path to production is a requirement for any company that wants to practice continuous deployment, because deployment to production ought to be a normal, everyday occurrence — not a rare, catastrophic, all-hands-on-deck ceremony.
Rebecca Parsons (@rebeccaparsons), the CTO of Thoughtworks, showed us new ways for Operations to perform and why they need to evolve. For her, Operations, Continuous Delivery and evolutionary architecture need to work hand in hand. One of her most powerful statements was:
It’s irresponsible to use a microservices architecture if you don’t have the right and matured DevOps department and Continuous Delivery. It’s too risky.
She also emphasised the need for Operations to be like any other stakeholder in the company, like Marketing or Sales. To that end, it is important that Operations concerns be treated as first-class citizens.
Adrian Colyer (@adriancolyer) mostly focused on Operability through the lens of academia. On his blog called ‘The Morning Papers’ (https://blog.acolyer.org/) , he dissects the latest research and development in computer sciences. For this conference, he curated a special selection.
One the most interesting questions he asked was: In today’s world of microservices, how can we see the big picture? How can we know what is going on at every level? One of the tools that Facebook put in place is a ‘mystery machine’ that measures end-to-end performance in the system, from the loading of a page in a browser all the way to its backend infrastructure. But these types of solution create other problems: how can you host and search through gigantic logs? How would you determine which pieces of data are the most relevant?
Sabine Wojcieszak’s (@sabinebendizen) talk was quite the treat: she walked onstage dressed as a Jedi! I have to admit that the little geek inside of me cried a little. She managed to seriously demonstrate the similarities between DevOps (and developers in general) and Jedis, particularly in how they communicate. Her best statement to illustrate the whole concept was: “Communication is like dancing on the razors’ edge. You are only a foot away from the dark side.” Used correctly, communication makes you a powerful Jedi and enables your organisation to function at an optimal level, but used incorrectly, it takes you closer to the dark side and creates more problems along the way.
To summarise, we learnt so much from the speakers at this conference and enjoyed the chance to chat with others who were interested in DevOps. A big thank-you to Marco for generously sponsoring our tickets — we hope we can go again next year!
