The first day of Kubecon 2018 has been exciting and if you have not read the updates, you can find them here. The second day has some interesting tidbits on many areas I think are relevant for companies adopting the Kubernetes platform and scaling it.
One of the key highlights of the day was Sarah Wells’s keynote on migrating 150+ microservices to Kubernetes and the challenges and learnings from the migration. The talk had some really good learnings and I am eagerly waiting for the full talk to be available on Youtube. For now some key messages:
Another gem of a session was from Spotify, some really good points there around community and culture. This is another talk I would be eagerly waiting to watch when the video is out. I have heard & read a lot Spotify’s work culture and it would be good to learn from their experiences. For example, another quote from the Spotify’s David Zolotusky is :
Do not isolate yourself, we all are facing same or similar problems
In last few days, there have been announcements from all major cloud providers, for example, Google, Azure & AWS have all embraced the Open Service broker API. This will allow workloads not only to run across clouds but across different platforms such as CloudFoundry and Kubernetes. Projects such as Kubevirt/Nomad allow you to run and manage VMs as well on Kubernetes and there is a talk which demoed one such concept with Kubevirt, the code can be found here. I personally think we are going to live in a real hybrid workload world for some time. Think of running containers along with VMs, some serverless offering and container-based services such as ACI/fargate :)
One of the most interesting projects I have been following is the Kubeflow - running machine learning workloads on Kubernetes. There are commercial players like RiseML which also allow you to run ML on Kubernetes.
Nvidia’s plan for GPU support on Kubernetes also shows the amount of traction for ML workloads on Kubernetes.
One of the interesting dashboards put up live at Kubecon was the measurement of CO2 level around. Using IOT devices at edge and Kubernetes as a backend makes sense but using Kubernetes/Containers also on devices will make it promising way to deploy consistently across the stack. There was a session on edge computing with Kubernetes by Cisco and it will be interesting to watch when it comes out.
One deck I really liked on the level of detail was securing Kubernetes continuously. I also loved the sketch notes created by Dan Romlein for Kubecon 2018 and you should definitely check out the following:
Looking forward to the final day of Kubecon 2018 EU!