Last week, I had the pleasure of visiting Copenhagen for KubeCon + CloudNativeCon Europe 2018, the sixth KubeCon event and most popular yet.
The event is hosted by the Cloud Native Computing Foundation (CNCF). Membership has grown significantly in the last couple of years (this year saw over 4,000 attendees compared to 500 who attended the inaugural conference in San Francisco in 2015), leading them to announce during the opening keynote a much improved Interactive Landscape and Trail Map which are both great ways to get oriented with the many projects the foundation supports. Kubernetes is, of course, chief among them, and it became the first to “graduate” beyond CNCF Incubating status.
Stay current on your favourite topics
As a technologist in the financial services industry, it was a great event: three days to listen to and learn from many of the industry’s most forward-thinking firms.
Returning KubeCon speaker Oliver Beattie, Head of Engineering at London-based challenger bank Monzo, gave a fascinatingly open dissection of a Kubernetes-related outage they’d suffered in October, going into the kind of detail that led to most of the audience, similarly battled-scarred, nodding their heads in sympathy as he revealed the chain of events that had caused the problem.
This kind of openness was a common theme, particularly reflected in a humble presentation from Dave Zolotusky of Spotify (The Human Side of Cloud Native Infrastructure) in which he discussed overcoming his initial reservedness in seeking help from other firms, soon finding people willing to share information with him and his team (name-checking Morgan Stanley in the process). It gave some meaning to “community”, a word frequently associated with the open source movement but rarely illustrated so clearly.
My favourite of the keynotes was Switching Horses Midstream: The Challenge of Migrating 150+ Microservices to Kubernetes by Sarah Wells, Director of Operations and Reliability at the FT. She gave an insider’s view of an enterprise-scale transformation programme that was packed with insights, including thoughts on the hidden costs of parallel running and the migration of even relatively recent code. For example, teams might open an older project to make a simple change as part of the transformation, but feel compelled to make other improvements too, eating into project time. Any developer who has looked back at a code they wrote six months ago could sympathise; it was food for thought for project and development managers.
I couldn’t list highlights without mentioning a favourite of mine, Simon Wardley, who brought the house down during the closing keynote with a very polished and funny iteration of his talk on Value Chain Mapping (Crossing the River by Feeling the Stones). The humour appeals to the skeptic in all of us, but Wardley Maps are a useful tool, bringing a little more scientific rigour to some of the diagrams (such as maturity curves) that many of us use routinely.
Whilst Kubernetes took centre stage, Prometheus (a monitoring solution with time series database) got almost as much attention, being mentioned in most sessions I attended, usually paired with visualisation tool Grafana. Perhaps it will be the next project to graduate. Of special interest to me in my current role was Scalable Monitoring using Prometheus with Apache Spark in which a knowledgeable team from Red Hat described their success in applying Prometheus & Grafana to the tuning of Spark applications. I have not used Prometheus before, but I have had plenty of Spark applications run out memory and prove tricky to diagnose. I intend to look into Prometheus as soon as I can.
“DevOps” was barely mentioned by any of the speakers I saw. I suspect that amongst the CNCF community, and in a conference about containerisation, DevOps practices are assumed. Instead, “GitOps” (a name coined by Weaveworks for a development and deployment lifecycle centred on Git) had achieved some popularity. On the surface, it may seem nothing radical (Weave themselves note that the approach is “90% best practices and 10% cool new stuff”) but it seems to have struck a chord. Cloudbees’ James Strachan gave an extremely polished demo of Jenkins X which I spent the evening playing around with. It is also worth a deeper look.
Most sessions were filmed and are available on YouTube. I’ll be watching the ones that I missed.
Next for Citihub Consulting: my colleagues will soon report back from Microsoft Build, Seattle, I look forward to hearing what they picked up!