Skip to content
back to talks
podcast · DevOps Deflope · #62

Kubernetes: Too Complex?

September 13, 2026 · DevOps Deflope · #62

Alexander Polomodov and Alexander Kachmashev join DevOps Deflope to discuss how Kubernetes became the orchestration standard and why operating it grew more complex as its capabilities expanded. The episode explores engineers’ cognitive load, internal developer platforms, and the limits of AI agents’ access to a cluster. In preparation for the conversation, Alexander Polomodov wrote a longread on how complexity moves from Kubernetes into platforms and agents.

Preparation longreadWritten summary

Episode participants

  • Sergey Sizov · host

  • Pavel Okhlopkov · host

  • Vitaly Khabarov · host

  • Alexander Polomodov · guest

  • Alexander Kachmashev · guest · internal cloud tech lead · Tochka

Episode topics

  1. 01How did teams move from docker-compose and a few virtual machines to Kubernetes and its ecosystem?
  2. 02Why do additional abstraction layers increase engineers’ cognitive load?
  3. 03How do CNI and eBPF change networking in Kubernetes?
  4. 04Why do companies build internal developer platforms, and which complexity do those platforms absorb?
  5. 05What makes uncontrolled AI-assisted administration risky, and how should agents’ access to a cluster be limited?