Skip to main content

Pod Affinity

Updated Apr 07, 2022 ·

Overview​

Pod Affinity is similar to Node Affinity, but instead of constraining nodes, it ensures that Pods are scheduled based on the labels of existing Pods running on those nodes.

Key Differences:

  • Node Affinity: Ensures Pods are placed on specific nodes.
  • Pod Affinity: Ensures Pods are co-located on the same node as other Pods with matching labels.

Pod Affinity supports the same operators as Node Affinity:

  • In
  • NotIn
  • Exists
  • DoesNotExist
  • Gt
  • Lt

Considerations​

When using Pod Affinity, keep the following in mind:

  • More computationally expensive than Node Affinity.
  • Not recommended for large clusters.
  • Pod labels are namespaced, just like Pods themselves.

After evaluating the conditions, the topology key is used to decide which node the Pod will be scheduled on. This key typically refers to physical domains such as data centers, regions, or server racks.


Feedback