Due to its vast array of capabilities and applications, it can be challenging to pinpoint the exact essence of what Kubernetes does. Today we are asking questions about orchestration and runtimes and trying to see if we can agree on whether Kubernetes primarily does one or the other, or even something else. Kubernetes may rather be described as a platform for instance! In order to answer these questions, we look at what constitutes an orchestrator, thinking about management, workflow, and security across a network of machines. We also get into the nitty-gritty of runtimes and how Kubernetes can fit into this model too. The short answer to our initial question is that defining a platform, orchestrator or a runtime depends on your perspective and tasks and Kubernetes can fulfill any one of these buckets. We also look at other platforms, either past or present that might be compared to Kubernetes in certain areas and see what this might tell us about the definitions. Ultimately, we come away with the message that the exact way you slice what Kubernetes does, is not all-important. Rigid definitions might not serve us so well and rather our focus should be on an evolving understanding of these terms and the broadening horizon of what Kubernetes can achieve. For a really interesting meditation on how far we can take the Kube, be sure to join us today!
Follow us: https://twitter.com/thepodlets
Key Points From This Episode:
What defines an orchestrator? Different kinds of management, workflow, and security.
Considerations in a big company that go into licensing, security and desktop content.
Can we actually call Kubernetes and orchestrator or a runtime?
How managing things at scale increases the need for orchestration.
An argument for Kubernetes being considered an orchestrator and a runtime.
Understanding runtimes as part of the execution environment and not the entire environment.
How platforms, orchestration, and runtimes change positions according to perspective.
Remembering the 'container orchestration wars' between Mezos, Swarm, Nomad, and Kubernetes.
The effect of containerization and faster release cycles on the application of updates.
Instances that Kubernetes might not be used for orchestration currently.
The increasingly lower levels at which you can view orchestration and containers.
The great job that Kubetenes is able to do in the orchestration and automation layer.
How Kubernetes removes the need to reinvent everything over and over again.
Breaking down rigid definitions and allowing some space for movement.
“Obviously, orchestrator is a very big word, it means lots of things but as we’ve already described, it’s hard to fully encapsulate what orchestration means at a lower level.” — @mauilion [0:16:30]
“I wonder if there is any guidance or experiences we have with determining when you might need an orchestrator.” — @joshrosso [0:28:32]
“Sometimes there is an elemental over-automation some people don’t want all of these automation happening in the background.” — @opowero [0:29:19]
Links Mentioned in Today’s Episode:
Apache Airflow — https://airflow.apache.org/
SSCM — https://www.lynda.com/Microsoft-365-tutorials/SSCM-Office-Deployment-Tool/5030992/2805770-4.html
Ansible — https://www.ansible.com/
Docker — https://www.docker.com/
Joe Beda — https://www.vmware.com/latam/company/leadership/joe-beda.html
Jazz Improvisation over Orchestration — https://blog.heptio.com/core-kubernetes-jazz-improv-over-orchestration-a7903ea92ca?gi=5c729e924f6c
containerd — https://containerd.io/
We are joined by Ellen Körbes for this episode, where we focus on Kubernetes and its tooling. Ellen has a position at Tilt where she works in developer relations. Before Tilt, she was doing closely related kinds of work at Garden, a similar company! Both companies are directly related to working...
Running Kubernetes on conventional operating systems is time-consuming and labor-intensive. Today’s guests Andrew Rynhard and Timothy Gerla have engineered a product that attempts to provide a solution to this problem.