DNS failover with Route 53: SAA-C03 study notes and practice questions
What to know about DNS failover with Route 53 for the SAA-C03 exam, from the Design Resilient Architectures domain (26% of the exam guide), then 1 free question to test it. Part of the SAA-C03 study guide.
Route 53 questions describe how users should be sent to endpoints, and the routing policy is the answer. Health checks decide when an endpoint drops out.
Routing policies
Policy
Sends users to
Typical scenario
Simple
One record (one or several values)
A single resource, no health checks
Failover
The primary while healthy, otherwise the secondary
Active-passive DR across Regions
Latency
The Region with the lowest latency for the user
Global app running in several Regions
Weighted
Endpoints in proportion to weights you set
Blue/green or canary releases, A/B testing
Geolocation
An endpoint by the user’s continent, country or US state
Content or compliance by location
Geoproximity
By distance, adjustable with a bias
Shifting traffic between Regions gradually
Multivalue answer
Up to eight healthy records, chosen at random
Simple client-side load spreading with health checks
IP-based
An endpoint by the user’s IP range (CIDR)
Routing known networks, such as an ISP’s, to a specific endpoint
Key facts
Health checks can test an endpoint (HTTP, HTTPS or TCP), watch other health checks (calculated), or follow a CloudWatch alarm, which is how you check private resources Route 53 can’t reach.
Alias records point to AWS resources (load balancers, CloudFront, S3 websites, API Gateway, another record in the zone). They work at the zone apex (example.com), where a CNAME can’t, and Route 53 doesn’t charge for alias queries to AWS resources.
Evaluate target health on an alias record uses the target’s own health (for example a load balancer with no healthy targets) without a separate health check.
TTL controls how long resolvers cache an answer; a lower TTL means faster failover but more queries.
Common traps
Using a CNAME at the zone apex. Not allowed; use an alias.
Expecting DNS failover for a client that caches IP addresses or can’t change address. That’s a Global Accelerator scenario (static anycast IPs).
Latency vs geolocation. Latency is about speed; geolocation is about the user’s location, for rules or content.
A company runs its customer portal behind an Application Load Balancer in eu-west-2, with a smaller standby copy behind another load balancer in eu-west-1.
The standby must receive no user traffic while the primary is working. If every target behind the primary load balancer fails its health checks, users must be sent to the standby automatically, without anyone changing DNS records.
Which Amazon Route 53 configuration meets these requirements?
Choose one answer.
AA. Latency-based alias records for the two load balancers, with Evaluate Target Health set to Yes so unhealthy records are skipped.
BB. Failover alias records, with the eu-west-2 load balancer as primary with Evaluate Target Health set to No, and eu-west-1 as secondary.
CC. Failover alias records, with the eu-west-2 load balancer as primary with Evaluate Target Health set to Yes, and eu-west-1 as secondary.
DD. Weighted alias records with a 90/10 split between the eu-west-2 and eu-west-1 load balancers, with Evaluate Target Health on both.
Show the answer and explanation
Answer: C
A. Wrong: Latency routing is active-active: each user is sent to the healthy load balancer with the lowest latency, so some users would reach the standby during normal running. That breaks the rule that the standby receives no traffic. Failover routing keeps it idle. Learn more: Active-active and active-passive failover (opens in a new tab)
C. Right: Correct. Route 53 answers with the primary while it is healthy, so the standby gets no traffic. With Evaluate Target Health on, the load balancer counts as unhealthy when a target group has no healthy targets, and Route 53 then answers with the secondary. Learn more: Values specific for failover alias records (opens in a new tab)
D. Wrong: Weighted routing would fail away from an unhealthy primary, but the 10% weight sends part of normal traffic to the standby. That breaks the rule that the standby stays idle. Failover routing keeps all users on the primary until it is unhealthy. Learn more: Active-active and active-passive failover (opens in a new tab)
Explanation
Active-passive failover uses failover records: Route 53 returns the primary while it is healthy and the secondary only when it is not. For an alias to an Application Load Balancer, setting Evaluate Target Health to Yes makes Route 53 use the health of the load balancer's target groups. A record with no health check and no target health evaluation is always treated as healthy, so it never fails over.
With Evaluate Target Health on, Route 53 sees when every target behind the primary fails and starts answering with the standby.
Want more on DNS failover with Route 53? The 5 full-length practice sets have 10 more DNS failover with Route 53 questions, 325 in all, none of them in the free exam. One payment of $14.99, no subscription. See the practice sets →