5 take-aways from CloudNative- / KubeCon 2019

5 take-aways from CloudNative- / KubeCon 2019

5 take-aways from CloudNative- / KubeCon 2019 Barcelona about Kubernetes, Serverless and Cloud Native

Posted on 2019-06-04 by Frederik Denkens

In what is becoming a yearly fixed appointment, Skyscrapers attended CloudNative- / KubeCon 2019 which took place 2 weeks ago in the sunny Barcelona. If you are in any way involved with serverless, DevOps, containers and/or Kubernetes this is definitely a must-be-there event for you as well.

The adjoining expo provided a good current state of the ecosystem: monitoring & instrumentation is still a moving target, all major vendors hopped aboard and security is the next big thing. Those vendors also made sure we had plenty to drink and eat in the evenings at one of the many nice after-conference events.

The meat of the conference was however to be found in the great schedule.

Below we present to you our 5 main take-aways. It was written by Philip, Simon and Frederik who attended.

Insights image

Failure Stories

Who doesn’t love a good failure story? We definitely do! It might surprise you 😉 but using Kubernetes as a platform to run your applications doesn’t magically solve all problems (it does for some). It comes with its own set of caveats.

Therefore the most interesting user stories are when things go wrong. These “horror” stories provide valuable feedback and are a perfect way to learn from others running Kubernetes in production.

We enjoyed these failure story presentations:

For more stories, check out https://k8s.af/ (and maybe add your own story?).

Insights image

Grafana Loki

As often is the case with Kubernetes we standardize logging on an Elasticsearch, Fluentd and Kibana stack (EFK) for our customers’ managed clusters. However this turns out to be quite overkill and overly complex for some of our customers coupled in addition to the relatively high cost of running this stack.

Imagine our enthusiasm when following the Grafana Loki presentation at KubeCon. Loki promises a simple, cost effective and highly scalable log aggregation system inspired by Prometheus . Compared to Elasticsearch it doesn’t fully index the logs and logs are stored unstructured. Loki uses Kubernetes Pod labels, just like Prometheus, as indexed metadata instead. Furthermore it integrates with the Grafana dashboard as a new data source, so you can seamlessly switch between metrics and logs using the same Prometheus and Loki labels.

Loki is still in heavy development and are no “stable” releases available yet (first beta should land any moment). However we’ve already started implementing a proof-of-concept, using S3 as a chunk store and DynamoDB as an index store, to evaluate it as an alternative to our EFK based stack. (let us know if you want to test along with us)

The serverless landscape

The CNCF started in 2017 with the Serverless Working Group to explore what serverless means for the cloud. This landscape is now defined in their cloudnative landscape and they published the first serverless whitepaper .

Due to serverless being cloud specific the Serverless SIG started with CloudEvents, a specification for describing event data in a common way with the goal of making it easier to define your serverless code to be cloud-native.

The serverless eco-system is still in active development on all ends. With the addition of custom runtimes on AWS Lambda it’s now possible to run most of the programming languages serverless (yes even Cobol).

We also see that cloud providers invest heavily to make their platforms more serverless like AWS with Firecracker . MicroVMs like these in combination with frameworks like Knative allows container-based platforms to run serverless workloads.

The full in depth presentation can be found here

Insights image

The enormous force of Kubernetes & Cloud Native

Since getting involved about 3 years ago, it’s incredible to see the growth. Especially considering that Kubernetes is this year only turning 5!

Let’s start with some numbers for each of the KubeCon + CloudNativeCon Europe events:

  • 2017: 1.500 attendees (Berlin)
  • 2018: 4.300 attendees (Copenhagen)
  • 2019: 7.700 attendees (Barcelona)

Not just those numbers are impressive, as Cheryl Hungs keynote made clear:

  • Today 36 projects are tracked by the CNCF in which …
  • over 2.66M contributions have been made by …
  • 56.000+ contributors

Why is the community and market growth around this so strong? We think Dan Khon’s keynote summed it up quite nicely: ever growing community, no lock-in thanks to open source and a strong vendor-neutral strategy.

What does it mean for us? That we are not just following the hype, but are part of one of the greater things happening recently in the IT industry.

The human side of the CNCF

Tech can be interesting and cool, but it’s just that … technology. Hence it’s really nice to see a tech conference also open up to such topics as inclusiveness, diversity and mental health:

The human strategy behind the (CNCF) community strongly resonates with that of Skyscrapers. It’s a great feeling to recognise your own values in the larger community you rely on.

Conclusion

Other things we liked or found remarkable:

We thought it was great! In any case we will be integrating a lot of what we learnt into everything we do for our customers.

What should you do after reading this post?

  1. Head out to the YouTube playlist to see all sessions
  2. Save the date for next year: March 30 – April 2, 2020 | Amsterdam, The Netherlands
  3. Become a member of a local Kubernetes or Cloud Native meetup group.
  4. Join the Kubernetes and CNCF Slack organisations and become part of the community

And finally ... Contact us if you want to explore your Cloud Native journey together with one of our specialists.

Oh ... and want to join us next time? Feel free to join us as we're currently hiring!

Thanks for reading! Simon, Philip and Frederik.