Route 53 Routing Policies Explained: Simple, Weighted, Latency, Failover and More

How Amazon Route 53 decides which answer to give: simple, weighted, latency-based, failover, geolocation, geoproximity, multivalue and IP-based routing, plus health checks and alias records.

Amazon Route 53 is AWS's DNS service. Beyond simply translating names into IP addresses, it can choose which answer to return based on health, location, latency or weights. That makes DNS a powerful tool for high availability, gradual rollouts and global applications.

DNS basics in Route 53

  • A hosted zone holds the records for a domain. Public zones answer the internet; private zones answer only inside associated VPCs.
  • Records have a type (A, AAAA, CNAME, MX, TXT, NS…) and a TTL: how long resolvers cache the answer.
  • Alias records are a Route 53 extension that points a name at AWS resources (ALB, CloudFront, S3 website, API Gateway, another record). Unlike CNAMEs, they work at the zone apex (example.com), cost nothing to query for AWS targets and follow the target's changing IPs automatically.

The routing policies

PolicyReturnsUse for
SimpleThe record's value(s)A single resource
WeightedRecords in proportion to weightsCanary releases, A/B tests, gradual migrations
Latency-basedThe Region with the lowest latency for the userMulti-Region apps
FailoverPrimary while healthy, otherwise secondaryActive-passive disaster recovery
GeolocationRecord for the user's continent, country or stateLocalized content, legal restrictions
GeoproximityNearest resource, with adjustable biasShifting traffic between Regions gradually (traffic flow)
Multivalue answerUp to eight healthy records at randomSimple client-side load spreading
IP-basedRecord based on the client's IP rangeRouting specific networks or ISPs

Weighted routing for safe releases

Create two records with the same name and type, each with a weight. Weights 95 and 5 send roughly 5% of new lookups to the new environment. Increase gradually while watching error rates; set the old weight to 0 to finish. Because resolvers cache answers, keep TTLs low (for example 60 seconds) during migrations.

Latency-based routing

For applications deployed in several Regions, latency records send each user to the Region with the lowest measured network latency from their resolver. Combine with health checks so an unhealthy Region is skipped.

Failover routing and health checks

Failover routing needs a primary and a secondary record. Route 53 returns the primary while its health check passes and the secondary when it fails.

Health checks can monitor:

  • An endpoint over HTTP, HTTPS or TCP from checkers in several locations, optionally looking for a string in the response.
  • Other health checks (calculated checks combining several with AND/OR logic).
  • A CloudWatch alarm, useful for private resources that checkers cannot reach.

For alias records pointing at AWS resources such as an ALB, enable Evaluate target health instead of a separate health check. The secondary can be a static S3 "we'll be right back" page or a warm standby environment in another Region or AZ.

Geolocation vs geoproximity vs latency

  • Geolocation is about where the user is (for compliance or language). Always add a default record for unmatched locations.
  • Latency is about what is fastest.
  • Geoproximity is about distance, with a bias you can tune to expand or shrink a Region's area.

Private hosted zones and hybrid DNS

Private hosted zones give internal names like db.internal.example inside your VPCs. To resolve names between AWS and an on-premises network, use Route 53 Resolver inbound and outbound endpoints with forwarding rules, and protect outbound DNS with Resolver DNS Firewall.

Common pitfalls

  • High TTLs that make failover and migrations take hours to reach users.
  • Using CNAME at the zone apex (not allowed); use an alias record.
  • Health checks blocked by security groups or firewall rules on the endpoint.
  • Forgetting that DNS failover only affects new lookups; clients holding connections need application-level retries.

Practice it

Build active-passive failover in Route 53 Failover DR: Primary and Warm Standby. For internal and hybrid DNS, try Private Hosted Zones Across VPCs, Hybrid DNS with Route 53 Resolver and Route 53 Resolver DNS Firewall.

FAQ

Can I register domains in Route 53?

Yes. Route 53 is also a domain registrar, and it creates the hosted zone for you.

Is Route 53 a load balancer?

Not in the traditional sense. It influences which address clients connect to, but it does not see or balance individual connections. Use it with ALBs or NLBs, not instead of them.

Browse all AWS tutorials · Hands-on AWS labs