What are requests and limits for Pods? What are they used for and which ones should be set in production?

In Kubernetes, there are two important parameters for Pods — requests and limits, which specify CPU and memory resources.

• requests — the minimum guaranteed resources that Kubernetes will allocate to the Pod. The scheduler uses these values to determine on which node to place the Pod.
• limits — the maximum resources that the Pod can use. If the Pod exceeds the CPU limit, its usage will be throttled. If it exceeds memory, the Pod can be killed (OOMKilled).

Why they are needed:

• Ensure predictability and stability of the cluster.
• Prevent a situation where one Pod consumes all the node's resources.
• Help the scheduler distribute the load efficiently.

What to set in production:

• requests should reflect the minimum necessary for the stable operation of the application.
• limits — the maximum allowable resources to prevent excessive consumption.

Recommendations:

• Measure the actual needs of the application in a testing environment.
• Set requests slightly below the average consumption, limits — slightly above the peaks.
• Monitor metrics and adjust settings as needed.

Example:

resources:
  requests:
    cpu: "500m"
    memory: "256Mi"
  limits:
    cpu: "1"
    memory: "512Mi"

Thus, properly configured requests and limits ensure a balance between efficient resource utilization and application stability.