Why it matters
Teams often create one ALB per service or environment. The hourly charge of each adds up even when traffic is near zero, while traffic charges move to the shared ALB and stay about the same.How nOps finds it
nOps looks for ALBs billed on each of the last 14 days that carried traffic but averaged under 0.5 LCU an hour over the last 7 days, where at least 3 such ALBs are in the same account and region. They are grouped in blocks of up to 10 (ALBs have limits on rules and certificates), the busiest ALB in each block is kept, and the saving is the hourly charge of the others.Before you start
The ALBs must share a VPC, compatible security groups and certificates.How to implement
- Confirm the ALBs to merge are in the same VPC and have compatible security groups, certificates and settings.
- On the ALB being kept, add listener rules (host or path based) that forward to each merged ALB’s target groups, and add their certificates.
- Point each DNS record at the kept ALB and wait for traffic to move.
- When the merged ALB shows no traffic, delete it.
- For ALBs created by the AWS Load Balancer Controller in Kubernetes, use an IngressGroup (
group.nameannotation) instead of changing the ALBs by hand.
Risks and considerations
- An ALB supports up to 100 rules and 25 certificates per listener.
- A shared ALB is a shared failure point and shares security settings, so keep separate ALBs where isolation matters.
- Never delete an ALB that Kubernetes created: change the Ingress or Service instead, or the controller recreates it.
FAQs
Does traffic cost change?
Does traffic cost change?
LCU charges move to the kept ALB with the traffic and stay about the same. The saving is the hourly charge of each ALB removed.