Key takeaways from the Madrid DevOps conference. That is where the K8s all gathered around their pods with a Helm. If you are a cloud native speaker, these things should be crystal clear.
Perspective: moving from a classic VM oriented architecture to container based app runtime infrastructure.
Complicated decisions on cloud native resources.
- Need for a lot of diff tooling for networking, load balancing, monitoring… fragmented Kubernetes landscape still a challenge. Hence importance abstraction via cloud services hiding kubernetes platform complexity.
https://www.influxdata.com/blog/will-kubernetes-collapse-under-the-weight-of-its-complexity/ - When you try to run your own kubernetes cluster in production, innocent kitties will die 😉
- Managing a Kubernetes platform requires a dedicated team
- That team needs appropriate training and experience
https://www.sdxcentral.com/articles/news/kubernetes-security-plagued-by-human-error-misconfigs/2020/02/ - Your scale needs to justify the overhead of setting up your own cluster, vs leveraging Amazon AWS, Google CS to use a managed Kubernetes environment.
https://www.weave.works/technologies/the-journey-to-kubernetes/#gke-gcp
- choose the right cloud provider, MS Azure, IBM, Google Cloud, Oracle (avoid vendor lockin?)
> Hybrid solution?
Overcome Vendor Lock-In by Integrating Already Available Container Technologies ( Towards Transferability in Cloud Computing for SMEs) - Move from classic devops to container based devops; Requires careful restructuring of teams and extra cognitive load for devs.
- Very close Dev-Ops communication, has advantage of immediate feedback loop when deploying changes to prod.
- Still importance for a dedicated Ops role for Monitoring, CI/CD tooling, QA… Having one team that handles all is not an optimal strategy in this sense.
Source: Team topologies - Organizations … understand the link between organizational structure and the software they create. For example, Netflix and Amazon structure themselves around multiple small teams, each one autonomous over a small part of the system.
Perspective: CI tools SaaS vs on premise tools
- TCO on premise CI is high:on premise ci hosting eg. Jenkins may have a low initial cost but maintenance hours add up quickly over time (broken plugins, cost of jenkins server itself etc)
- See slides on top of article with comparison charts of features.
https://medium.com/@rothgar/why-kubernetes-is-abbreviated-k8s-905289405a3c
! Hasta luego Madrid ✌️