These are some notes and takeaways on the recently celebrated Kafka Summit 2018 in London.
The conference was organized in three parallel tracks for sessions that were covering stream processing, pipelines, and internals. To get a good experience, I attended talks of the three types, but with a little preference towards internals and streams.
It was a two-day conference with lots of valuable technical content, awesome talks, speakers, and a lot more. Here are the highlights.
The first keynote, “The Death and Rebirth of the Event Driven Architecture”, was the perfect way to start the conference. Totally a mind-opening talk, to get the right idea of what is the state of the art in terms of distributed systems, and event-driven architectures in modern software engineering.
Some of the main ideas, phrases, and concepts of this talk:
Not just the architectures, but also the business are event-oriented. You can think a business, every business, as a large set of events occurring at all times. The architecture then, is just reflecting this.
Every business is now digital. Start-ups create their architecture, and build their business upon it. Legacy companies, adapt to this, so their software architecture becomes the core of the business, and they operate based on this.
The rest of the talks were amazing as well, and in the majority I chose talks of the tracks of stream processing and pipelines with some exceptions on which I attended a talk about internals that explained key metrics to monitor in a Kafka installation, how Kafka uses the JVM heap internals, and why this (along with a lot of more metrics, such as the Kernel I/O pagination cache) are key metrics to monitor.
At the end of the day, I had the chance to talk to some of the speakers and keep up enlightening discussions about event sourcing architectures.
Shorter than the previous one (finished at around 1pm), but the talk were equally amazing.
It was impressive to see how some big companies use Kafka directly on Docker containers, with Kubernetes, and a monitoring platform automatically sending metrics to Grafana. There was a talk explaining the immense infrastructure at CERN, and how Kafka plays a crucial role on their data processing solutions.
All in all it was a great conference, on which I learnt a lot about Kafka, distributed systems, event sourcing architecture. Besides learning about the technology it was great to see how companies do real-world implementation of these solutions, and exchange opinions with fellow professionals.