Explanation
Why the waste happens and who it affects.
It only does anything useful when the same name has latency records in two or more Regions. When a name has a single latency record, every query is answered with that one Region, yet Route 53 still bills the query at the higher latency-based routing rate instead of the standard rate. The AWS Trusted Advisor check for this case states that all queries then go to one Region and you pay extra for latency-based routing without getting the benefits.
This usually happens when teams create latency records in anticipation of a multi-Region rollout that never happens, when a second Region is decommissioned but the remaining record keeps its routing policy, or when templates apply latency routing by default. The premium matters only for non-alias records in public hosted zones with high query volumes, because alias queries to supported AWS resources and queries to private hosted zones are not charged.
Billing model
The pricing dimensions that drive this cost.
Route 53 bills public hosted zone DNS queries per million, at a rate set by the routing policy of the record that answers.
- Standard queries
- Billed per million queries for simple, failover, multivalue answer and weighted records, $0.40 per million for the first 1 billion a month in AWS Regions
- Latency-based routing queries
- Billed per million queries for latency records, $0.60 per million for the first 1 billion a month in AWS Regions
- Alias queries
- Free when an alias A or AAAA record targets a supported AWS resource such as an ELB, CloudFront distribution or API Gateway
- Private hosted zone queries
- Not charged
How to detect
4 checks to find it in your estate.
- Review the AWS Trusted Advisor cost optimization check Amazon Route 53 Latency Resource Record Sets, which flags names with only one latency resource record set
- List records with ListResourceRecordSets and group latency records (those with a Region and SetIdentifier) by name and type to find names with a single Region
- Exclude alias records that target supported AWS resources, since their queries are free regardless of routing policy, and focus on non-alias A, AAAA and CNAME latency records
- Use the Route 53 usage report or Cost Explorer to size latency-based routing query volume for the affected public hosted zones
How to fix
4 ways to remove the waste.
- If the workload runs in only one Region and there is no multi-Region plan, replace the latency record with a simple record
- Where the target is an AWS resource that supports alias records, use a simple alias record so the queries are not charged at all
- If multi-Region serving is intended, add latency records for each Region that hosts the workload so the routing policy delivers its benefit
- Lower the record TTL before the change and restore it afterwards to shorten the transition, and update any IaC templates that apply latency routing by default
Documentation
Vendor references for pricing and configuration.