[#546] Trochę o Kubernetesie na produkcji - Michał Schott

Key Concepts and Recommendations:

• Kubernetes Basics:
  • Kubernetes consists of components like the API server, controller manager, etcd, and worker nodes responsible for managing pods.
  • The speaker suggests using containerd instead of Docker for container management, as it is lighter and more efficient.
• Cluster Management:
  • There are two approaches to setting up Kubernetes clusters: self-managed and managed services. Each has its pros and cons, but managed services can simplify operations.
  • The importance of understanding the cluster lifecycle and the potential issues with etcd (e.g., desynchronization due to IP changes) is highlighted.
• Operating Systems for Kubernetes:
  • Lightweight operating systems (e.g., Ubuntu, Debian) are recommended for running containers without unnecessary overhead.
• Scaling and Cost Management:
  • The speaker emphasizes the need for strategies like node TTL (time-to-live) to manage resources efficiently and avoid memory leaks.
  • Recommendations include using cluster autoscalers and ensuring that workloads are balanced across nodes to minimize costs.
• Networking and Security:
  • Cilium is recommended as a CNI (Container Network Interface) for its visibility and support for network policies.
  • The speaker discusses the importance of service meshes for traffic encryption and management.
• Operators and Automation:
  • Operators can automate the management of applications within Kubernetes, but debugging them can be challenging.
  • The use of Helm charts for deploying applications is also discussed, with a preference for reviewing what is being deployed.
• Resource Management:
  • The speaker discusses the importance of setting resource limits for CPU and memory and the implications of not doing so.
  • Vertical and horizontal scaling strategies are outlined, with a recommendation for using Horizontal Pod Autoscalers (HPA).
• Disruption Budgets:
  • The concept of Pod Disruption Budgets (PDB) is explained as a way to manage the availability of applications during updates or scaling events.
• Best Practices:
  • The importance of using proven solutions, establishing update processes for worker nodes, and monitoring traffic for optimization are emphasized.

Tools and Techniques Mentioned:

• Certifications: Recommendations for obtaining CKA (Certified Kubernetes Administrator) and CKS (Certified Kubernetes Security Specialist) certifications.
• Monitoring: Use of tools like Prometheus for metrics and Hubble for real-time traffic observability.
• Security Tools: Mention of tools like Falco and Alma Security for runtime protection.

Main Speaker:

• Michał Schott: The primary speaker sharing his insights and experiences with Kubernetes over the past years.

Overall, the video serves as a comprehensive guide for those looking to deepen their understanding of Kubernetes in production environments, offering practical advice and insights based on real-world experiences.