Back to all scenarios
Scenario #425
Scaling & Load
Kubernetes v1.27, AWS EKS

Inadequate Load Distribution in a Multi-AZ Setup

Load balancing wasn’t even across availability zones, leading to inefficient scaling.

Find this helpful?
What Happened

More traffic hit one availability zone (AZ), causing scaling delays in the other AZs.

Diagnosis Steps
  • 1Analyzed kubectl describe svc and found skewed traffic distribution.
  • 2Observed insufficient pod presence in multiple AZs.
Root Cause

The Kubernetes service didn’t properly distribute traffic across AZs.

Fix/Workaround
• Updated service to use topologySpreadConstraints for better AZ distribution.
Lessons Learned

Multi-AZ distribution requires proper spread constraints for effective scaling.

How to Avoid
  • 1Use topologySpreadConstraints in services to ensure balanced load.
  • 2Review multi-AZ architecture for traffic efficiency.
Previous ScenarioNext Scenario