One decision question for each routing policy
Simple
Do I have one resource or a straightforward set of values without special routing logic?
Weighted
What relative proportion of DNS answers should point to each resource?
Latency
Which configured AWS Region offers the requester the lowest measured latency?
Failover
Should a secondary resource answer only when the primary is unhealthy?
Geolocation
What answer should users receive based on their geographic location?
Geoproximity
How should traffic shift among resource locations, optionally using bias?
IP-based
Do known client CIDR ranges require specific answers?
Multivalue answer
Should DNS return up to eight randomly selected healthy records?
Scenario-based choices
Release 5% of users to a new version
Create weighted records with the same name and type. If the stable endpoint has weight 95 and the new endpoint has weight 5, the new endpoint receives a 5/100 share of Route 53 selections over time.
Record share = record weight ÷ total weight
Multi-Region application optimized for response time
Create latency records for the application endpoint in each deployed AWS Region. Route 53 evaluates its latency data and returns the record for the Region expected to provide the best latency for that requester.
Primary Region with a warm standby Region
Use failover records with primary and secondary roles and health evaluation for the primary. This encodes active-passive intent directly; weighted routing alone does not express “use secondary only when primary is unhealthy.”
Country-specific compliance endpoint
Use geolocation when the user’s location determines the answer. Add a default record so requesters whose location does not match a configured rule still receive a valid response.
Health checks change which answers are eligible
Route 53 health checks can monitor an endpoint, the status of other health checks, or a CloudWatch alarm. DNS failover can then omit unhealthy resources and return a healthy alternative. For supported alias records, evaluate-target-health can use the health of the referenced AWS resource.
DNS is cached: failover is not an instantaneous connection migration. Resolvers and clients may continue using an answer until its TTL expires, and existing connections are not moved by changing a DNS response.
Design the health signal carefully. A shallow check can report success while the application cannot serve users; an overly strict check can remove a healthy endpoint because of one noncritical dependency.
Common SAA-C03 mistakes
- Choosing geolocation when the requirement says “lowest latency.”
- Choosing latency routing when a business rule requires users from a specific country to use a specific endpoint.
- Treating weighted routing as exact per-request traffic splitting despite DNS caching.
- Using weighted records for active-passive recovery instead of failover records.
- Claiming multivalue routing provides all the capabilities of an Elastic Load Balancer.
- Forgetting a default geolocation record for locations that do not match a configured rule.
- Assuming a low TTL eliminates every cache or existing-connection delay.
Fast decision rule
Percentage: weighted. Performance: latency. Primary/secondary: failover. User location: geolocation. Resource geography plus traffic bias: geoproximity. Known client CIDR: IP-based. Several healthy answers: multivalue.
Official AWS references
Frequently asked questions
What is the difference between weighted and latency routing?
Weighted routing divides DNS responses according to relative weights that you configure. Latency routing selects among configured AWS Regions according to Route 53 latency measurements for the requester.
Which Route 53 policy supports active-passive disaster recovery?
Failover routing uses primary and secondary records, normally with health evaluation, to support an active-passive design.
Is multivalue answer routing a replacement for an Elastic Load Balancer?
No. Multivalue routing can return up to eight healthy records selected at random, but it does not provide the connection-level traffic distribution and features of a load balancer.
What is the difference between geolocation and geoproximity?
Geolocation routes based on the location of users. Geoproximity considers the location of resources and can use bias to shift traffic toward or away from a resource location.