The original title is: "Kubernetes resource requests and limits explained: scheduling, throttling, and OOMKill"

Short summary

Kubernetes resource requests and limits serve different purposes that prevent one of the most common production failures: requests tell the scheduler how much memory to reserve for placement decisions, while limits are hard runtime ceilings that kill pods exceeding them. The critical insight is that the scheduler uses requested resources for placement, not actual usage, creating a balancing act between cost efficiency and system stability. Getting this right prevents the 3am OOMKilled cascades.

• •Requests tell the scheduler about resource needs; limits enforce hard runtime ceilings
• •Scheduler uses requests (not actual usage) for placement decisions
• •Misaligned requests/limits cause OOMKilled cascades, one of the most preventable production incidents

Generated with AI, which can make mistakes.