Route 53 & DNS
AWS SAA SAA-C03 · Design Resilient Architectures
Route 53: DNS in the cloud
Amazon Route 53 is AWS's Domain Name System (DNS) service. It translates human-readable domain names (like example.com) into IP addresses, registers domain names, and — most importantly for the exam — routes users to AWS resources using health-aware routing policies. Route 53 is a global service with 100% availability as its design target, and it is the exam's answer for almost any question involving domain names, DNS routing, or DNS-based failover.
A hosted zone is a container for DNS records for a domain. A public hosted zone holds records that are resolvable on the internet; a private hosted zone holds records that resolve only inside specified VPCs, which is how you give internal services friendly names like db.internal without exposing them publicly. Records map names to values: A records map a name to an IPv4 address, AAAA to IPv6, CNAME maps a name to another name, and NS and SOA records are created automatically for the zone. Route 53 also supports alias records, which are a Route 53 extension: an alias record maps a name directly to an AWS resource — an ELB, a CloudFront distribution, an S3 website endpoint, or another record in the same hosted zone — with no charge for the queries and automatic tracking of the target's changing IP addresses. Alias records can be used at the zone apex (the bare domain, example.com), where CNAME records are not allowed. When an exam question asks how to point a bare domain at a load balancer or CloudFront distribution, the answer is an alias record.
Routing policies: how Route 53 chooses an answer
| Policy | Behavior | Typical use |
|---|---|---|
| Simple | Returns one value (or multiple in random order) | A single resource; no health logic needed |
| Weighted | Splits traffic by assigned weights | Blue/green deployments, A/B testing, gradual migration |
| Latency-based | Routes to the Region with the lowest latency for the user | Multi-Region applications where responsiveness matters |
| Failover | Routes to a primary resource, switching to a secondary on health check failure | Active-passive disaster recovery |
| Geolocation | Routes based on the user's geographic location | Localized content, regulatory requirements, content licensing |
| Geoproximity | Routes based on geographic distance, with adjustable bias | Shifting traffic between Regions by distance |
| Multi-value answer | Returns multiple healthy values | Simple load balancing across healthy resources |
The exam tests these policies by describing a requirement and asking which policy fits. The distinctions that matter most: weighted is for controlled traffic splits (a percentage to the new version); latency-based is for performance (nearest Region wins); geolocation is for location-based rules (users in Europe see the European site); failover is for disaster recovery (primary/standby). Note that latency-based routing measures latency between the user and AWS Regions, not geographic distance — a user can be geographically closer to one Region but have lower latency to another.
Health checks and failover
Route 53 health checks monitor the health of an endpoint — an IP address, a domain, or a CloudWatch alarm — by sending requests at intervals and checking the response. Health checks are what make failover and multi-value routing work: Route 53 only returns healthy endpoints. A failover configuration has a primary record and a secondary record; when the health check on the primary fails, Route 53 starts answering with the secondary. Health checks can also be based on CloudWatch alarms, which lets you fail over based on any metric — for example, failing over when the primary Region's error rate crosses a threshold, even if the endpoint still responds to pings. Health checks can be configured with a failure threshold (how many consecutive failures before marking unhealthy) and can monitor endpoints outside AWS too.
One subtlety worth knowing: DNS-based failover is not instant. Clients and resolvers cache DNS answers for the record's TTL, so some users may keep hitting the failed endpoint until their cached answer expires. A short TTL (such as 60 seconds) makes failover faster but increases query volume. This is why DNS failover is a good fit for disaster recovery measured in minutes, while Global Accelerator — which fails over at the network layer without DNS changes — is the better fit when failover must be nearly instant.
DNS in hybrid and private architectures
Route 53 also handles DNS for hybrid networks. Route 53 Resolver provides DNS resolution inside a VPC, including inbound endpoints (so on-premises systems can resolve names in your VPC) and outbound endpoints (so VPC resources can resolve on-premises names). Resolver rules let you forward queries for specific domains to your own DNS servers. For multi-account architectures, Route 53 Resolver endpoints can be shared across accounts, and private hosted zones can be associated with VPCs in other accounts. The exam rarely goes deep here, but the concept — Route 53 Resolver connects VPC DNS with on-premises DNS — is worth knowing.
Advanced Route 53: Resolver, Traffic Flow, and health checks
Beyond basic routing policies, Route 53 offers tools for complex and hybrid environments. Route 53 Resolver handles DNS resolution inside a VPC and bridges it with on-premises DNS: inbound endpoints let on-premises systems resolve names hosted in your VPC's private hosted zones, while outbound endpoints let VPC resources forward queries for specific domains to your own DNS servers. Resolver rules define which domains are forwarded where, and they can be shared across accounts in an organization so every VPC uses the same forwarding logic. Route 53 Resolver DNS Firewall adds a protective layer, blocking outbound DNS queries to known malicious domains and letting you define your own allow and block lists — useful for stopping malware that uses DNS to reach command-and-control servers.
Route 53 Traffic Flow is a visual editor for building complex routing configurations that combine multiple policies — for example, geolocation routing with failover inside each location, or weighted routing across latency-based endpoints. It also supports versioning, so you can roll a routing change back if it misbehaves. Health checks have depth worth knowing: a calculated health check combines the results of other health checks (for example, healthy if at least two of three endpoints respond), and a CloudWatch alarm health check lets Route 53 fail over based on any metric, such as an application error rate, rather than endpoint reachability alone. Health checks can monitor endpoints outside AWS, and you can configure the failure threshold (how many consecutive failures mark an endpoint unhealthy) and the request interval. Finally, Route 53 can register and manage domain names directly, and domain transfer lets you move an existing domain's registration to Route 53 without downtime.
One final practical detail: Route 53 charges a small monthly fee per hosted zone and per million DNS queries, and alias queries to AWS resources are free. Health checks are billed per check per month, so a design with many health checks across many endpoints should account for that cost — though it is usually small compared with the resources being monitored.
Exam traps to watch for
- Using a CNAME record at the zone apex. CNAMEs cannot be used for the bare domain (example.com) — use an alias record pointing at the load balancer or CloudFront distribution.
- Confusing latency-based routing with geolocation routing: latency-based routes to the lowest-latency Region; geolocation routes by the user's location.
- Expecting instant failover from DNS. Cached answers mean some users keep hitting the failed endpoint until the TTL expires — use a short TTL, or Global Accelerator for near-instant failover.
- Choosing a simple routing policy when the scenario needs health checks or traffic splitting. Simple routing has no health awareness.
Practice questions
Test yourself on this guide. Answers are checked instantly, with a full explanation for each question.
Q1.A company wants its bare domain (example.com, with no subdomain) to point to an Application Load Balancer. The load balancer's IP addresses change over time. Which Route 53 record type should the solutions architect use?
Explanation: An alias A record maps the zone apex directly to the load balancer and automatically tracks its changing IP addresses. It is the only option that works at the apex for an AWS resource. Option B is invalid: CNAME records cannot be used at the zone apex. Option A would break whenever the load balancer's IPs change, which happens routinely. Option C is wrong: NS records delegate DNS authority for a zone, they do not point a domain at a load balancer.
Exam objective: SAA-C03 Task Statement 3.4: Determine high-performing and/or scalable network architectures
Q2.A company is deploying a new version of its website and wants to send 10% of user traffic to the new version and 90% to the current version, gradually increasing the share if the new version performs well. Which Route 53 routing policy should the solutions architect use?
Explanation: Weighted routing splits traffic between records according to assigned weights — exactly the mechanism for a controlled 10/90 rollout that can be adjusted over time. Option A routes by lowest latency, which has nothing to do with traffic percentages. Option B routes by the user's geographic location. Option C is active-passive: all traffic goes to the primary until it fails, which is a disaster-recovery pattern, not a gradual rollout.
Exam objective: SAA-C03 Task Statement 2.2: Design highly available and/or fault-tolerant architectures
Q3.A company runs its application in two Regions: a primary Region and a standby Region. If the primary Region becomes unavailable, users must be routed to the standby Region automatically. Which Route 53 configuration meets this requirement?
Explanation: Failover routing with health checks is the active-passive disaster recovery pattern: Route 53 monitors the primary endpoint and starts answering with the secondary record when the health check fails. Option D has no health awareness — Route 53 would keep returning the failed endpoint. Option B routes by latency and would send users to whichever Region is faster, not maintain a primary/standby arrangement. Option C routes by geographic distance with a bias, which is for traffic shaping, not failover.
Exam objective: SAA-C03 Task Statement 2.2: Design highly available and/or fault-tolerant architectures
Q4.A global application is deployed in three Regions. The company wants each user to be routed to the Region that will give them the lowest network latency. Which Route 53 routing policy should be used?
Explanation: Latency-based routing measures latency between the user and each AWS Region and routes to the lowest-latency one — precisely the requirement. Option B routes by the user's geographic location, which does not always match lowest latency (a user can be geographically near one Region but have lower latency to another). Option C splits traffic by fixed percentages. Option A returns multiple healthy endpoints for the client to choose from but does not select by latency.
Exam objective: SAA-C03 Task Statement 3.4: Determine high-performing and/or scalable network architectures
Q5.A company must show different website content to users based on the country they are connecting from, to comply with regional content licensing rules. Which Route 53 routing policy meets this requirement?
Explanation: Geolocation routing returns different answers based on the geographic location (continent, country, or US state) of the user, which is exactly what content licensing by country requires. Option C routes by network latency, ignoring location. Option B is for primary/standby failover. Option D returns the same answer to everyone.
Exam objective: SAA-C03 Task Statement 3.4: Determine high-performing and/or scalable network architectures
Q6.A company has internal microservices running in a VPC. Developers want to reach the services by friendly DNS names (for example, orders.internal) that resolve only inside the VPC and are not visible on the public internet. Which solution meets this requirement?
Explanation: A private hosted zone holds DNS records that resolve only within the VPCs associated with it, keeping internal names off the public internet. Option D publishes the records publicly, which is the opposite of the requirement. Option B also exposes the names publicly. Option A exposes the services themselves to the internet, which is a security failure.
Exam objective: SAA-C03 Task Statement 1.2: Design secure workloads and applications
Q7.A solutions architect configures Route 53 failover between a primary and a secondary website. During a test, some users continue to reach the failed primary site for several minutes after failover. What is the MOST likely explanation?
Explanation: DNS answers are cached by resolvers and clients for the record's TTL, so users with a cached answer keep hitting the primary until that cache expires — even after Route 53 has switched to the secondary. A shorter TTL reduces this window. Option A is backwards: a shorter health check interval makes failover faster, not slower. Option C is irrelevant to DNS caching. Option B is wrong: failover routing is automatic once health checks fail.
Exam objective: SAA-C03 Task Statement 2.2: Design highly available and/or fault-tolerant architectures
Q8.A company wants Route 53 to fail over to a secondary Region when the primary Region's application error rate exceeds a threshold, even if the primary endpoint still responds to health check pings. Which type of Route 53 health check supports this?
Explanation: Route 53 health checks can be based on the state of a CloudWatch alarm, so failover can be triggered by any metric — such as an application error rate — rather than only by endpoint reachability. Option D only checks whether the endpoint responds, which it still does in this scenario. Option C monitors the wrong endpoint. Option A is a health check that combines other health checks; without a CloudWatch alarm in the mix it cannot react to the error-rate metric.
Exam objective: SAA-C03 Task Statement 2.2: Design highly available and/or fault-tolerant architectures