Route 53 hosts Domain Name System (DNS) records, the entries that let clients look up a domain name. Routing policies and health signals determine which answers it returns. VPC Resolver looks up public and private names for virtual private clouds (VPCs); endpoints and forwarding rules connect those lookups to DNS servers on other networks.
Route 53 basics
Amazon Route 53 is AWS’s DNSA system that translates domain names into records such as IP addresses. service: it registers domains, answers DNS queries and checks endpointA network address and port that receive queries or health probes. health.
How it works
- Register a domain or use an existing one.
- Create DNS recordsAn entry for a name and type that supplies values in a DNS answer. pointing to resources.
- Health checksA configured test or health signal used to determine whether a resource is healthy. steer traffic away from failed endpoints.
Exam cues
- If the question says DNS or domain registration, think Route 53.
Remember
- Domain registration, DNS routing, health checks.
Trap: Route 53 answers DNS queries; it does not carry the application traffic.
From the AWS docs (7)
- Amazon Route 53 is a highly available and scalable DNS web service for domain registration, DNS routing and health checking.
- Failover routing uses a primary resource and a standby secondary for active-passive operation.
- Active-active configurations use policies other than failover and use record health to exclude unavailable resources.
- Route 53 chooses a record using the routing policy, checks its current health status, and tries another if it is unhealthy; probes run separately from DNS queries.
- Weighted records with the same name and type receive DNS answers in proportion to their weight divided by the total weight.
- Route 53 latency routing compares configured Regions using measurements collected over time between users and AWS data centers. An endpoint outside AWS can have substantially different latency, even when it is near an AWS Region.
- Route 53 health checks monitor the health of web applications and other resources, and CloudWatch alarms can notify when a health check status changes.
Public zone delegation
A public hosted zoneA container of DNS records for a domain. holds DNS recordsAn entry for a name and type that supplies values in a DNS answer. for a domain. Domain registration and DNSA system that translates domain names into records such as IP addresses. hosting can use different providers. A registrarThe provider that manages registration of a domain name. manages the domain registration; delegationThe assignment of DNS authority to name servers for a domain. tells DNS which name servers answer for it.
How it works
- If the domain is registered elsewhere, create a public hosted zone in Route 53.
- At the registrar, configure the four name servers assigned to that hosted zone. These are its authoritative servers, meaning they supply its DNS answers.
- Add the records needed for the application. Creating a zone without updating delegation does not make the registrar use it.
- When registering through Route 53, Route 53 creates the zone and updates the domain’s name servers for you.
| Record | Value it provides |
|---|---|
| A | IPv4 address |
| AAAA | IPv6 address |
| CNAMEA DNS record that names another domain as its destination. | Another domain name; not allowed at the zone apexThe DNS name that is the same as the hosted-zone name. |
| NS | Authoritative name serversA server that supplies the DNS records for its hosted zone. |
| MX | Mail servers and preference values |
| TXT | Text, including domain ownership values |
Exam cues
- If the question says keep the current registrar and use Route 53 DNS, think update name-server delegation, because registration transfer is unnecessary.
- If the question says the new zone has correct records but public DNS uses old answers from another provider, think delegation, because the registrar must point to the intended zone’s servers.
Remember
- Route 53 assigns four name servers to the hosted zone.
- A uses IPv4; AAAA uses IPv6.
Trap: Creating a public hosted zone for an externally registered domain does not update that registrar’s name-server settings.
Create the zone and obtain its assigned name servers.
Configure the registrar to use those servers so the zone supplies DNS answers.
From the AWS docs (12)
- A domain can keep its existing registrar while using Route 53 for DNS hosting.
- For a domain registered elsewhere, create a Route 53 public hosted zone and configure the registrar to use the four name servers assigned to that zone.
- Registering a domain with Route 53 creates its hosted zone, assigns four name servers and updates the domain to use them.
- An A record provides an IPv4 address.
- An AAAA record provides an IPv6 address.
- A CNAME maps a name to another domain name and cannot exist at the hosted-zone apex.
- NS records identify the authoritative name servers for a hosted zone.
- MX records specify mail servers and their preference values.
- TXT records hold text values, including values used to establish domain ownership.
- Route 53 DNSSEC signing signs every response for a hosted zone with public key cryptography, so DNS resolvers can verify that answers came from Route 53 and weren't tampered with.
- To move an active domain to Route 53 DNS, recreate the records in a hosted zone, lower the TTLs, wait for the old TTL to expire, then point the domain NS records to the Route 53 name servers.
- A shorter DNS TTL makes resolvers notice changes sooner but raises query volume and cost; a longer TTL serves more answers from cache but delays changes. AWS recommends 60 to 172,800 seconds.
Routing policies
Route 53 selects DNSA system that translates domain names into records such as IP addresses. answers using routing policies. DNS, the Domain Name System, translates a name into recordAn entry for a name and type that supplies values in a DNS answer. values such as IP addresses. The policy chooses the answer; health signals can change which resources are eligible.
How it works
- Create records for the application name and choose the policy that matches the routing requirement.
- For weighted routing, give records the same name and type. Each record’s share is its weight divided by the total weight.
- Route 53 applies the policy and checks the selected record’s current health. EndpointA network address and port that receive queries or health probes. probes run independently of queries.
- A recursive resolverA DNS server that looks up an answer for a client and can cache the result., the server that looks up names for a client, caches the answer for its time to liveThe period for which a DNS resolver can cache a record answer. (TTL). Cached answers can serve many application requests.
| Policy | Selection rule |
|---|---|
| Simple | One resource for a domain function |
| Failover | Primary and standby |
| Weighted | Relative share of DNS answers |
| Latency | AWS Region selected using measured latency |
| Geolocation | User continent, country or US state |
| Geoproximity | Distance between user and resources, adjusted by biasA geoproximity setting that expands or shrinks the geographic area served by a resource. |
| IP-based | User IP ranges in a CIDRAn IP address range written using an address and a prefix length. collection |
| Multivalue answer | Up to eight healthy records |
Latency routing uses measurements to AWS data centers, which may differ from actual latency to endpoints outside AWS.
A bias expands or shrinks the geographic area served by a geoproximity resource. A CIDR block is a compact way to describe an IP address range.
Exam cues
- If the question says gradually move DNS answers to a new stack, think weighted routing, because weights control relative answer shares.
- If the question says route users by country, think geolocation with a default record, because unmapped locations otherwise receive no answer.
- If the question says shift traffic gradually between two environments with DNS, think weighted records for two separate endpoints, because each environment needs its own load balancer to weight.
Remember
- A simple record returns all its values; multivalue records can check each resource.
- The smallest matching geolocation wins.
- An unchecked record counts as healthy.
- Weighted with health checksA configured test or health signal used to determine whether a resource is healthy.: zero-weight records answer only when every nonzero record is unhealthy. If every record is unhealthy, Route 53 treats all as healthy.
- Multivalue answers are not a substitute for a load balancer.
- Geolocation can localize content or enforce distribution rights.
Trap: Weighted DNS does not promise an exact percentage of application requests. Resolvers can serve a cached answer repeatedly.
Select a regional branch using latency data.
Use weighted selection among healthy endpoints.
If Region A has no healthy leaves, try Region B.
Return a value from the healthy regional branch.
From the AWS docs (17)
- A routing policy controls which record value Route 53 returns for a DNS query.
- Simple routing is for a single resource performing a function for a domain.
- Weighted records with the same name and type receive DNS answers in proportion to their weight divided by the total weight.
- Recursive resolvers cache answers for the record TTL, so a cached answer can serve requests without another Route 53 query.
- Geolocation routing uses user location by continent, country or US state; the smallest overlapping location takes precedence.
- A default geolocation record covers unmapped addresses and locations without a matching record; without it, those queries get no answer.
- Geoproximity routing uses user and resource locations; a positive bias expands the area served by a resource and a negative bias shrinks it.
- IP-based routing maps user IP address ranges, called CIDR blocks, to chosen endpoints through CIDR collections.
- Multivalue answer routing returns up to eight healthy records selected at random, or all healthy records when there are eight or fewer.
- A record without an associated health check is treated as healthy.
- For a blue/green shift with Route 53, each environment exposes its own endpoint, such as its own load balancer, and weighted records send a percentage of traffic to green, raised step by step until green carries all traffic.
- A simple record with several values returns all of them in random order and the resolver hands them to the client; unlike multivalue answer records, the values are not health checked one by one.
- Geoproximity routing locates AWS resources by Region or Local Zone group and non-AWS resources by latitude and longitude.
- With health checks on weighted records, Route 53 first considers only nonzero-weight records and uses zero-weight records only when all nonzero-weight records are unhealthy.
- When no record in a group is healthy, Route 53 treats all of them as healthy and still answers according to the routing policy.
- Multivalue answer routing returns health-checked IP addresses to improve uptime and share load, but it is not a substitute for a load balancer.
- Geolocation routing can serve content in the user's language or limit content to the locations where you have distribution rights.
Endpoint and calculated health checks
Route 53 health checksA configured test or health signal used to determine whether a resource is healthy. supply a health signal for DNSA system that translates domain names into records such as IP addresses. routing. An endpointA network address and port that receive queries or health probes. check probes an address and port. A calculated check combines component results so DNS can represent application health.
How it works
- Choose an endpoint, child health checks, or a CloudWatch alarm data stream to monitor. An endpoint is the network address and port receiving the probe.
- For an endpoint, allow the checker addresses through routing and firewall rules. A security group can use the AWS-managed health-check prefix lists.
- For a calculated check, count healthy children and compare the count with the required threshold. Calculated checks cannot be children of another calculated check.
- Attach the check to the recordAn entry for a name and type that supplies values in a DNS answer. whose selection should depend on it. The check continues probing its configured target independently of DNS queries.
Exam cues
- If the question says users can connect but checkers cannot, think firewall and routing access for checker addresses, because probes must reach the configured port.
- If the question says every application component must be healthy, think a calculated check with an all-children threshold, because it combines child results.
Remember
- The health check probes its own configured target.
Trap: Attaching a check to a record does not make the check probe the address in that record. It probes its own configured target.
From the AWS docs (9)
- Endpoint checks send HTTP, HTTPS or TCP requests to the endpoint configured in the health check.
- TCP-only checks must establish a connection within ten seconds.
- The health-check settings support monitoring endpoints, child health checks, or a CloudWatch alarm data stream.
- A calculated check compares the count of healthy children with a threshold; the console offers all, one or more, or at least a chosen count.
- A calculated health check cannot monitor another calculated health check.
- Associating a health check with a record does not change the endpoint being probed; it uses the endpoint specified in that check.
- Routing and firewall rules must allow checker source addresses to reach the configured endpoint port.
- Security groups can allow the AWS-managed com.amazonaws.<region>.route53-healthchecks prefix list and the ipv6.route53-healthchecks list for IPv6 endpoints.
- Endpoint probes run periodically rather than when a DNS query arrives.
Alarm-based health checks
An alarm-based Route 53 health checkA configured test or health signal used to determine whether a resource is healthy. turns a CloudWatch metricA series of measurements that an alarm can evaluate. signal into DNSA system that translates domain names into records such as IP addresses. health. It is useful for a private application that public endpointA network address and port that receive queries or health probes. checkers cannot reach. A metric is a series of measurements; an alarm evaluates those measurements against a condition.
How it works
- Choose a standard-resolution metric that represents application health and configure a CloudWatch alarm in the same account.
- Identify that alarm by its name and Region when creating the Route 53 health check.
- Route 53 reads the alarm’s metric data stream. A result corresponding to OK is healthy; ALARM is unhealthy.
- Choose how the check handles insufficient data. Attach the check to the DNS recordAn entry for a name and type that supplies values in a DNS answer. whose routing should depend on that health signal.
Exam cues
- If the question says private IPs cannot be reached by Route 53 checkers, think an alarm-based check, because health can come from a metric rather than a public probe.
Remember
- The alarm must be in the same account. Its identifier includes its Region.
- Supported statistics: Average, Minimum, Maximum, Sum and SampleCount.
- High-resolution metrics, extended statistics, M-out-of-N alarms and multi-metric math alarms are unsupported.
- Insufficient data can mean Healthy, Unhealthy or LastKnownStatus. New LastKnownStatus checks start healthy.
Trap: A Route 53 endpoint health check cannot probe a private IP address. For a resource reachable only inside a VPC, base the check on a CloudWatch alarm instead.
Represent private application health with a metric evaluated by an alarm.
Use the alarm metric data stream as the Route 53 health signal.
From the AWS docs (9)
- Endpoint health checks cannot probe local, private, non-routable or multicast IP address ranges.
- An alarm-based health check reads the CloudWatch alarm metric data stream; OK means healthy and ALARM means unhealthy.
- The CloudWatch alarm must be in the same account as the health check.
- AlarmIdentifier identifies a CloudWatch alarm by its name and AWS Region.
- Alarm-based checks support standard-resolution metrics, not high-resolution metrics.
- Supported alarm statistics are Average, Minimum, Maximum, Sum and SampleCount; extended statistics are unsupported.
- Alarm-based checks do not support M-out-of-N alarms or alarms using metric math to query multiple metrics.
- InsufficientDataHealthStatus can be Healthy, Unhealthy or LastKnownStatus; a new check using LastKnownStatus begins healthy.
- SetAlarmState cannot force a Route 53 health-check change because Route 53 reads the metric stream.
Alias records and target health
A Route 53 aliasA Route 53 record that returns values from a supported AWS target or a same-type record in the same hosted zone. points a DNSA system that translates domain names into records such as IP addresses. name at a supported AWS resource or a same-type recordAn entry for a name and type that supplies values in a DNS answer. in the same hosted zoneA container of DNS records for a domain.. It lets a root domainThe DNS name that is the same as the hosted-zone name. use a changing AWS endpointA network address and port that receive queries or health probes. without storing that endpoint’s IP addresses yourself.
How it works
- Choose the supported target and an appropriate record type, such as A for an IPv4 answer or AAAA for an IPv6 answer. The zone apex is the hosted-zone name itself.
- Route 53 answers with the target’s values. It uses the AWS resource TTLThe period for which a DNS resolver can cache a record answer., or the target record TTL for an alias to another record.
- Enable Evaluate Target Health when inherited target health fits the requirement. For an Application Load Balancer (ALB) or Network Load Balancer (NLB), it uses target groupsA collection of destinations whose health a load balancer monitors., the collections of destinations behind the load balancer.
- If you also attach a separate check to the alias, both health conditions must pass.
For fixed network entry addresses across Regions, a Route 53 alias can point to a standard Global Accelerator. The accelerator selects healthy regional endpoints using its routing settings. It terminates client TCPA network protocol that establishes a reliable connection between endpoints. connections at AWS edge locations and opens connections to the endpoints. TCP is a protocol for reliable network connections.
Exam cues
- If the question says the root domain must point to an ALB, think an alias, because a CNAMEA DNS record that names another domain as its destination. cannot exist at the zone apex.
- If the question says an ALB alias needs a health signal, think Evaluate Target Health, because it uses the load balancer target groups.
- If the question says fixed entry IPs and healthy endpoints across Regions, think an alias to Global Accelerator, because the accelerator handles network routing behind those addresses.
Remember
- AWS-resource alias queries have no Route 53 query charge.
- You cannot set a custom TTL for an alias to an AWS resource.
Trap: An alias cannot point directly to an EC2 instance public DNS name. Choose a supported target.
From the AWS docs (13)
- An alias returns values for a supported AWS target or another record of the same type in the same hosted zone.
- An alias can exist at the zone apex, the hosted-zone name itself; a CNAME cannot.
- An alias to an AWS resource uses that resource TTL; an alias to another record uses that record TTL.
- Route 53 does not charge for alias DNS queries to AWS resources.
- EvaluateTargetHealth makes an alias inherit the health of its referenced target.
- For an ALB or NLB to be healthy, every associated target group containing targets needs at least one healthy target.
- An associated target group without registered targets is considered unhealthy.
- Use load-balancer health with Evaluate Target Health rather than separate Route 53 probes of the registered instances.
- If an alias has its own health check and Evaluate Target Health, both health conditions must pass.
- A Route 53 alias can route an application hostname to a Global Accelerator.
- A standard Global Accelerator routes through fixed entry-point IP addresses to healthy regional endpoints using location, health and routing settings.
- Global Accelerator terminates client TCP connections at an AWS edge location and establishes TCP connections to application endpoints.
- Alias targets are API Gateway APIs, VPC interface endpoints, CloudFront, App Runner, Elastic Beanstalk, Elastic Load Balancing, S3 website buckets, OpenSearch Service, Global Accelerator or another record in the same zone; an EC2 instance is reached with an A record to its Elastic IP.
DNS failover and testing
Route 53 failover routing changes the DNSA system that translates domain names into records such as IP addresses. answer from a primary resource to a standby. It uses recorded health status, while recursive resolverA DNS server that looks up an answer for a client and can cache the result. caches control when a client receives a fresh answer.
How it works
- Configure primary and secondary failover recordsAn entry for a name and type that supplies values in a DNS answer. for the application name, with health checksA configured test or health signal used to determine whether a resource is healthy. or applicable aliasA Route 53 record that returns values from a supported AWS target or a same-type record in the same hosted zone. target health.
- With checks on both records, Route 53 returns the primary while it is healthy. It returns the secondary when the primary is unhealthy and the secondary is healthy.
- An unchecked secondary is returned whenever the primary fails, regardless of the secondary resource’s actual condition.
- A resolver can continue returning an earlier answer until its TTLThe period for which a DNS resolver can cache a record answer. expires. TTL means the time an answer may stay cached.
Exam cues
- If the question says the secondary should answer only after a primary failure, think failover records, because the routing policy implements active-passive selection.
- If the question says resolvers still return the primary after its check fails, think TTL, because a cached answer avoids another Route 53 lookup.
- If the question says show an S3 error page after the primary ALB fails, think a primary ALB alias and a secondary S3 alias, because failover routing selects the standby after the primary becomes unhealthy.
Remember
- AWS recommends a TTL of 60 seconds or less for health-checked records.
- Probes run separately from DNS queries.
- Control plane (record and health-check changes): us-east-1. Data plane (answers, health checks): global.
Trap: An unchecked secondary is treated as healthy even if its resource is unavailable.
The resolver asks Route 53 when its cached answer has expired.
While the primary is healthy, Route 53 answers with the primary record.
If the primary is unhealthy and the secondary is healthy, return the secondary. If both checked records are unhealthy, return the primary.
Box details (4)
- Client
A resolver can reuse a cached DNS answer until its TTL expires.
- Route 53
With both records checked, return the healthy primary, otherwise the healthy secondary. If both fail, return the primary.
- Primary endpoint
Primary failover record using a health check or applicable alias target health.
- Secondary
Secondary failover record. An unchecked secondary is treated as healthy.
From the AWS docs (8)
- With health checks on both failover records, Route 53 returns the primary while it is healthy.
- When the checked primary is unhealthy and checked secondary is healthy, Route 53 returns the secondary.
- A secondary without a health check is returned whenever the primary is unhealthy, even if the secondary resource has failed.
- Resolvers can keep returning a cached answer until its TTL expires.
- For records associated with health checks, AWS recommends a TTL of 60 seconds or less.
- DNS queries use existing health status rather than starting a probe.
- Active-passive failover uses the failover routing policy: Route 53 answers with the healthy primary resources and only switches to the secondary resources when every primary is unhealthy.
- For Route 53 public and private DNS and health checks, the control plane, which handles record and health-check changes, is in us-east-1, while the data planes that answer queries and run health checks are globally distributed.
Resolver DNS Firewall
Route 53 ResolverA DNS server that looks up an answer for a client and can cache the result. DNSA system that translates domain names into records such as IP addresses. Firewall controls which domain names clients can resolve through VPC Resolver, the DNS service used by a virtual private cloud (VPC). DNS translates names into recordsAn entry for a name and type that supplies values in a DNS answer. such as addresses. DNS Firewall filters those queries; it does not inspect HTTPS payloads or turn domain names into blocked destination IPs.
How it works
- Create a domain listA collection of domain specifications used to match DNS queries., a collection of names to match, or select an AWS-managed list. Foundational rules can also select a query type. DNS Firewall Advanced detects DNS tunnelingThe use of DNS queries to carry data outside the normal purpose of name resolution., which carries data through DNS, and DGAA method attackers use to generate many domain names for malicious activity. threats, which generate domain names for malicious use.
- Add rules to a rule groupA reusable collection of DNS Firewall rules associated with one or more VPCs., a reusable rule collection, and associate the group with a VPC. Filtering also covers on-premises queries entering that protected VPC through Resolver endpointsA network address and port that receive queries or health probes..
- Choose an action for each rule.
ALLOWpermits the query and stops inspection;BLOCKstops inspection and returns the configured block response. Queries with no match resolve normally.
For an allow-only policy, allow approved domains before a rule blocking the rest. Actions · Policy examples
By default, domain-list rules inspect all names in a DNS redirection chainThe sequence of names followed when DNS records point a lookup to other names., where one name points to another. A setting that trusts redirected names applies only to that query; a direct query for the destination is evaluated independently.
Exam cues
- If the question says block malicious domain lookups through Amazon-provided DNS, think DNS Firewall, because security groups and network ACLs cannot block that resolver path.
- If the question says an allow list must deny everything else, think a later catch-all block rule, because unmatched queries resolve normally.
- If the question says deploy common DNS controls across accounts, think RAM sharing or Firewall Manager policies, because rule groups can be reused across VPCs. Sharing · Organization policies
Remember
- Queries that match no rule resolve normally.
- Advanced rules support
BLOCKandALERT, notALLOW.
Trap: A matching allow rule stops inspection, so a later block rule does not inspect that query.
Trap: Blocking a DNS name does not inspect or deny all traffic to its destination. DNS filtering and application-traffic controls solve different problems.
Send a name query through VPC Resolver.
Evaluate associated groups and rules from lowest numeric priority.
Apply the first matching action; if none matches, resolve normally.
From the AWS docs (14)
- DNS Firewall filters domain-name queries through VPC Resolver, including queries from VPC resources and on-premises queries entering the protected VPC through Resolver endpoints.
- DNS Firewall does not resolve domain names into blocked IP addresses and does not filter other application protocols such as HTTPS.
- A rule group is a reusable collection of filtering rules; associating it with a VPC enables filtering for that VPC.
- A query that matches no rule across the associated rule groups resolves normally.
- Foundational rules can use user-managed or AWS-managed domain lists and can optionally restrict a rule to a DNS query type.
- By default, domain-list rules inspect the entire DNS redirection chain; trusting redirected domains applies only within that query transaction.
- ALLOW stops inspection and permits the query; it is not available for DNS Firewall Advanced rules.
- BLOCK stops inspection and blocks the intended DNS destination, using the configured DNS block response.
- An allow-only design evaluates approved domains before a rule blocking the remaining domains.
- AWS RAM can share rule groups across accounts; recipients can associate them with their VPCs but cannot edit or delete the shared group.
- Firewall Manager DNS Firewall policies manage rule-group associations across an AWS organization.
- Security groups cannot filter traffic to or from Amazon-provided DNS.
- Network ACLs cannot block traffic to or from the Route 53 VPC Resolver.
- DNS Firewall Advanced detects DNS tunneling and DGA threats, including dictionary DGAs, and supports BLOCK or ALERT actions.
Public DNS query logging
Public DNSA system that translates domain names into records such as IP addresses. query logging recordsAn entry for a name and type that supplies values in a DNS answer. the queries that reach a Route 53 public hosted zoneA container of DNS records for a domain.. It helps inspect requested names and DNS response codes, which describe whether resolution succeeded or returned an error.
How it works
- Choose or create a CloudWatch Logs log groupA CloudWatch Logs container for related log streams. in us-east-1 for the hosted zone’s query logs. A log group contains related log streamsA sequence of log events from a particular source..
- Grant Route 53 permission to write through a CloudWatch Logs resource policyA policy attached to a resource that grants a principal access to it., a policy attached to the logging destination.
- Enable query logging for the public hosted zone. Route 53 publishes a separate log stream for each responding edge location.
- Inspect the requested name, type, timestamp, location and response code in CloudWatch Logs. A resolverA DNS server that looks up an answer for a client and can cache the result. that answers from cache sends no query to Route 53 for that lookup.
Exam cues
- If the question says inspect public DNS names and response codes, think public query logging to CloudWatch Logs, because it records queries Route 53 receives.
- If the question says logs omit many client lookups, think resolver caching, because cached answers do not require another authoritative query.
Remember
- Public DNS query-log groups must be in us-east-1.
- Route 53 needs permission to publish to the log group.
- Each responding edge location has a separate log stream.
Trap: Public query logs are not a record of every client lookup or application request. Resolver caches can answer without contacting Route 53.
From the AWS docs (6)
- Public DNS query logging sends the queries received by Route 53 to CloudWatch Logs.
- The CloudWatch Logs log group for public DNS query logs must be in us-east-1.
- A CloudWatch Logs resource policy must allow Route 53 to publish to the selected log group.
- Public query logs include the requested name, record type, timestamp, responding edge location and response code.
- Public query logs record resolver queries reaching Route 53; queries answered from resolver caches do not reach Route 53.
- Route 53 creates a log stream for each edge location responding for the configured hosted zone.
Cross-account DNS sharing
Cross-account DNSA system that translates domain names into records such as IP addresses. can reuse central forwarding rules or let another account’s VPC resolve a private hosted zoneA container of DNS records for a domain.. These are separate operations: a shared rule forwards queries, while a zone association makes private recordsAn entry for a name and type that supplies values in a DNS answer. available to a VPC.
How it works
- To reuse forwarding, share the ResolverA DNS server that looks up an answer for a client and can cache the result. rule through AWS Resource Access Manager (RAM), the service for sharing supported AWS resources between accounts.
- The recipient associates the rule with its VPCs in the Region. It can use the rule but cannot edit or delete it.
- For a direct private-zone association, the zone owner authorizes each VPC with CreateVPCAssociationAuthorization.
- The VPC owner calls AssociateVPCWithHostedZone. This direct cross-account workflow uses an API, CLI or SDK, because the console cannot perform it.
- After association, delete the authorization if it is no longer needed. The association remains. Set both VPC DNS attributes to true.
Exam cues
- If the question says many accounts need the same forwarding domain, think share the Resolver rule through RAM, because each recipient can associate it with its VPCs.
- If the question says a VPC in another account needs records in a private zone, think authorize then associate, because the zone owner and VPC owner each perform a step.
Remember
- Shared Resolver rules are Regional.
- Create one direct zone-association authorization per VPC.
- Deleting an authorization does not delete its completed association.
Trap: Sharing a forwarding ruleA DNS rule that sends queries for a chosen domain to specified DNS servers. does not directly associate a VPC with a private hosted zone.
The zone owner creates a VPC association authorization.
The VPC owner completes the association; later authorization deletion leaves it intact.
From the AWS docs (9)
- Resolver forwarding rules can be shared with accounts through AWS Resource Access Manager (RAM).
- A recipient associates a shared Resolver rule with its VPCs to use it.
- A recipient cannot change or delete a shared Resolver rule.
- Resolver rules and their VPC associations are Regional resources.
- For direct cross-account private-zone association, the zone owner calls CreateVPCAssociationAuthorization for each VPC.
- After authorization, the VPC owner calls AssociateVPCWithHostedZone.
- The Route 53 console cannot authorize or create this direct cross-account private-zone association.
- Deleting the authorization after association does not remove the existing association.
- The VPC must have enableDnsSupport and enableDnsHostnames set to true for private-zone resolution.
Health and DNS operations
Route 53 health checksA configured test or health signal used to determine whether a resource is healthy., routing settings and ResolverA DNS server that looks up an answer for a client and can cache the result. DNSA system that translates domain names into records such as IP addresses. Firewall rules determine how DNS answers and queries behave during failures. Their operational details matter when a health signal, recordAn entry for a name and type that supplies values in a DNS answer. weight or firewall rule produces an unexpected result.
How it works
- EndpointA network address and port that receive queries or health probes. checkers use a 30-second or 10-second interval and a consecutive success or failure thresholdThe number of consecutive passing or failing probes needed to change an endpoint checker status.. Route 53 considers the endpoint healthy when more than 18% of checkers report healthy. Check settings · Health determination
- HTTP and HTTPS checks need a TCPA network protocol that establishes a reliable connection between endpoints. connection within four seconds and a 2xx or 3xx response within two seconds after connection. A string check needs the body within two more seconds and the complete string in its first 5,120 bytes. HTTPS checks do not validate TLSThe protocol that protects an HTTPS connection. certificates.
- For an ALB or NLB aliasA Route 53 record that returns values from a supported AWS target or a same-type record in the same hosted zone., each associated target groupA collection of destinations whose health a load balancer monitors. with targets needs at least one healthy target. An empty group is unhealthy. For S3, API Gateway and interface endpoints, AWS recommends checking application access for failover. CloudFront aliases cannot use Evaluate Target Health or exist in private hosted zones. Target health · CloudFront target
- For regional failover, latency aliases can point to regional weighted groups. Enable Evaluate Target Health and check the non-alias leaves: an unchecked leaf can keep a failed branch eligible. If both checked failover records are unhealthy, Route 53 returns the primary. Regional trees · Unchecked leaves · Failover selection
- DNS Firewall evaluates rule groupA reusable collection of DNS Firewall rules associated with one or more VPCs. associations and rules from the lowest numeric priorityThe numeric ordering setting that determines when a rule or rule group is evaluated. upward. An
ALERTmatch permits the query, records an alert in configured VPC Resolver query logs and stops inspection. Use it to test a proposed block. Configure logging for the VPC to CloudWatch Logs, S3 or Firehose in the logging configuration Region. Rule order · Actions · Logging
DNS Firewall block responses differ:
| Response | Result |
|---|---|
NODATA |
Successful query with no answer data |
NXDOMAIN |
The domain does not exist |
OVERRIDE |
A custom CNAMEA DNS record that names another domain as its destination., with a default TTLThe period for which a DNS resolver can cache a record answer. of zero |
TTL is the time a record can be cached.
Exam cues
- If the question says test primary failover while the endpoint is healthy, think an inverted health check, because inversion reverses the result. Disabling a non-inverted check makes it healthy. Check settings · Disabled checks
- If the question says an ALB alias is unhealthy despite healthy targets elsewhere, think an empty associated target group, because an empty group is unhealthy.
- If the question says a protected VPC must keep resolving when DNS Firewall does not respond, think
FirewallFailOpen, because the default fail-closed mode returnsSERVFAIL. - If the question says a domain rule misses subdomains, think a wildcard entry such as
*.example.com, because the parent entry does not include all subdomains. - If the question says all nonzero weighted records are unhealthy, think zero-weight records, because Route 53 considers them when health checks are configured.
Remember
- The 10-second check interval costs extra. The health-check console can create a CloudWatch alarm that notifies an SNS topic in us-east-1 after one minute of unhealthy status. Intervals · Notifications
- Weighted records accept integer weights from 0 through 255. If all weights are zero, resources have equal selection probability.
- Multivalue routing returns up to eight healthy records, or all healthy records if there are eight or fewer. If all are unhealthy, it returns up to eight unhealthy records.
- An alarm-based health check follows the metricA series of measurements that an alarm can evaluate. data stream, not a state set with SetAlarmState; update the check after changing the alarm.
- The Route 53 SLAA policy defining a service availability commitment and eligibility for credits. covers authoritative DNS, not the API or console.
- DNSSEC signing: a customer managed asymmetric KMS key backs each key-signing key, Route 53 manages the zone-signing key, and a DS record in the parent zone completes the chain of trust.
Trap: A healthy HTTPS check does not prove that the TLS certificate is valid. An expired or invalid certificate can still pass.
From the AWS docs (37)
- CloudFront alias records are unsupported in a private hosted zone.
- An alias to CloudFront cannot set EvaluateTargetHealth to true.
- For an ALB or NLB to be healthy, every associated target group containing targets needs at least one healthy target.
- An associated target group without registered targets is considered unhealthy.
- For S3, API Gateway and interface endpoints, target health offers no operational benefit for application failover; AWS recommends Route 53 checks of application access.
- Inverted reverses a health check result; a healthy check becomes unhealthy.
- Disabling a check stops monitoring and makes it healthy unless its result is inverted.
- If both checked failover records are unhealthy, Route 53 returns the primary.
- Route 53 considers an endpoint healthy when more than 18% of checkers report healthy.
- The failure threshold is the number of consecutive successes or failures required to change a checker status.
- Each checker uses a request interval of 30 seconds or 10 seconds; the 10-second option costs extra.
- HTTP and HTTPS checks must establish a TCP connection within four seconds.
- HTTP and HTTPS checks need a 2xx or 3xx status within two seconds after connecting.
- String checks need the body within two more seconds after the status, with the full search string in the first 5,120 bytes.
- HTTPS checks do not validate the TLS certificate and can pass with an expired or invalid certificate.
- The health-check console can create a CloudWatch alarm that notifies an SNS topic in us-east-1 after the endpoint has been unhealthy for one minute.
- ALERT stops inspection, permits the query, and records an alert in configured VPC Resolver query logs.
- Rule group associations and rules within a group are evaluated from lowest numeric priority upward until a match or exhaustion.
- With an associated rule group, DNS Firewall defaults to fail closed: a missing firewall response makes Resolver block the query with SERVFAIL.
- Enabling FirewallFailOpen allows queries through when DNS Firewall does not respond.
- The NODATA block response reports a successful query without answer data.
- The NXDOMAIN block response reports that the queried domain does not exist.
- OVERRIDE returns a custom CNAME; its override TTL defaults to zero, so the override record is not cached by default.
- A domain entry does not implicitly include all subdomains; add a wildcard such as *.example.com to include those subdomains.
- A blocking rule can first be tested with ALERT to observe queries that would be blocked.
- Resolver query logging must be configured for the VPC; supported destinations include CloudWatch Logs, S3, and Firehose in the query logging configuration Region.
- Multivalue answer routing returns up to eight healthy records selected at random, or all healthy records when there are eight or fewer.
- If all multivalue records are unhealthy, Route 53 returns up to eight unhealthy records.
- When every record in a weighted group has weight 0, Route 53 gives the resources equal probability.
- Latency aliases can point to regional groups of weighted records; Evaluate Target Health lets Route 53 leave a Region with no healthy records and try another Region.
- Non-alias leaves in a health-aware alias tree need health checks; an unchecked leaf can keep a failed branch eligible.
- Weighted record weights are integers from 0 through 255.
- With health checks on weighted records, Route 53 considers zero-weight records when all nonzero-weight records are unhealthy.
- SetAlarmState cannot force a Route 53 health-check change because Route 53 reads the metric stream.
- After changing the CloudWatch alarm settings, update its Route 53 health check.
- The Route 53 SLA covers authoritative DNS hosted-zone availability and excludes API and console availability.
- DNSSEC signing works only for public hosted zones; each key-signing key is backed by an asymmetric customer managed KMS key you own, Route 53 manages the zone-signing key, and a DS record in the parent zone completes the chain of trust.
Flashcards
Nothing to review here right now. Pick other topics or another filter.
·
All 69 cards as a list
A company needs to register a domain and route its traffic to resources on AWS. Which service fits?
Amazon Route 53.
An application runs in two AWS Regions, and its standby should receive DNS answers only after the primary fails. Which Route 53 policy fits?
Failover routing. It implements active-passive selection using primary and secondary records.
How does Route 53 support active-active failover?
Use a policy such as weighted or latency with health checks or alias target health. Healthy resources remain eligible for DNS answers.
A team wants a gradual DNS migration to a new application stack using Route 53. Which policy should it use?
Weighted routing. Give the stacks relative weights and increase the new stack weight as the migration proceeds.
Which Route 53 policy selects a Region using measured network latency?
Latency routing. It uses AWS latency data collected over time for the Regions with configured records.
Which Amazon Route 53 feature monitors whether a web server is healthy so DNS can fail over away from it?
Route 53 health checks.
A company keeps its domain at an external registrar but wants Route 53 DNS hosting. Must it transfer registration?
No. Create a public hosted zone and set the registrar to use the four Route 53 name servers for that zone.
Records in a new Route 53 public zone do not affect public lookups for an externally registered domain. What should be checked?
The registrar name-server configuration. It must delegate to the four name servers assigned to the intended hosted zone.
Which Route 53 DNS record types represent IP addresses, another name, name servers, mail servers and text?
A: IPv4. AAAA: IPv6. CNAME: another name. NS: authoritative name servers. MX: mail servers. TXT: text values.
A bank must protect its public domain against DNS spoofing so resolvers can verify Route 53 answers. Which Route 53 feature fits?
DNSSEC signing on the public hosted zone.
How do you move DNS for a live domain to Route 53 with little disruption?
Recreate the records in a hosted zone, lower TTLs, wait for the old TTL to expire, then change the NS records at the registrar to the Route 53 name servers.
How do Route 53 geolocation, geoproximity and IP-based routing differ?
Geolocation selects by user location. Geoproximity selects by user and resource locations with an optional bias. IP-based routing uses user IP ranges that you supply.
Why should Route 53 geolocation routing include a default record?
It covers unmapped IP addresses and locations without a specific record. Without a default, those queries receive no answer.
A team changes Route 53 weights and expects each application request to follow the new ratio immediately. Why can that fail?
Resolvers reuse cached answers until their TTL expires. Weights control DNS answer selection; cached answers can serve many application requests.
A team moves traffic gradually from an Auto Scaling group behind an ALB to a new Auto Scaling group using Route 53. How should the new environment be exposed?
Behind its own ALB, with weighted alias records for both load balancers. Raise the green weight in steps until it takes all traffic.
A simple record lists three IP addresses and one server dies. Does Route 53 stop returning its address?
No. A simple record returns every value in random order without checking each one; multivalue answer records with health checks return only healthy addresses.
Amazon Route 53 weighted records have health checks, and a zero-weight record points to a maintenance page. When does Route 53 return the zero-weight record?
Only when all nonzero-weight records are unhealthy.
Every record in an Amazon Route 53 weighted group fails its health check. Does Route 53 stop answering DNS queries for the name?
No. It treats all the records as healthy and answers by the routing policy.
A team wants Amazon Route 53 multivalue answers to replace its Application Load Balancer. Is that a good fit?
No. Multivalue answers improve uptime with health-checked IP addresses but are not a substitute for a load balancer.
A video service may stream some titles only in licensed countries. Which Amazon Route 53 routing policy helps?
Geolocation routing. It can limit content to the locations where you have distribution rights.
Which monitoring choices does the Route 53 health-check console provide?
An endpoint, the status of other health checks through a calculated check, or a CloudWatch alarm data stream.
How does a Route 53 calculated health check determine health?
It compares the healthy-child count with a threshold. Select all children, one or more, or a chosen count. It cannot monitor another calculated check.
Does attaching a Route 53 health check to a DNS record make it probe the address in that record?
No. It probes the endpoint configured in the health check. The association uses that result to influence record selection.
An application answers user requests, but Route 53 endpoint health checks cannot connect. What network configuration should be checked?
Allow the Route 53 health checker address ranges to reach the configured endpoint port. A security group can use the regional AWS-managed health-check prefix lists.
A team needs Route 53 to declare an application healthy only when every component check passes. What should it configure?
A calculated health check with the all-children threshold. Use endpoint or alarm checks as children, because calculated checks cannot be nested.
A private application needs Route 53 failover, but public checkers cannot reach it. How should health be represented?
Use a CloudWatch alarm-based health check on an application-health metric. Endpoint checks cannot probe private IP ranges.
Which CloudWatch alarms can back a Route 53 health check?
Same-account alarms identified by name and Region, using standard-resolution metrics and Average, Minimum, Maximum, Sum or SampleCount. Extended statistics, M-out-of-N alarms and multi-metric math alarms are unsupported.
What does a Route 53 alarm-based health check report when CloudWatch lacks enough data?
The configured Healthy, Unhealthy or LastKnownStatus result. A new LastKnownStatus check begins healthy. SetAlarmState cannot force a change.
What does Evaluate Target Health use for a Route 53 alias to an ALB or NLB?
Load-balancer target-group health. Each target group needs a healthy target; an empty group is unhealthy. Separate Route 53 probes of registered instances are unnecessary.
A root domain must point to an ALB through Route 53. Which record should be used?
An alias A record, or AAAA for an IPv6 answer. A CNAME cannot exist at the zone apex. The alias inherits its AWS target TTL, and AWS-resource alias queries have no Route 53 query charge.
A team needs a custom hostname and fixed network entry IPs for healthy application endpoints across Regions. What AWS combination fits?
A Route 53 alias to a standard Global Accelerator. The accelerator routes traffic through its fixed IP entry points to healthy regional endpoints.
Can a Route 53 alias record point at an EC2 instance public DNS name?
No. EC2 instances are not alias targets; use an A record with the instance Elastic IP.
Does changing Route 53 failover health immediately change every client DNS answer?
No. Resolvers can keep returning an earlier cached answer until its TTL expires.
A Route 53 primary fails, and the secondary has no health check. Is Route 53 checking whether the secondary works before returning it?
No. It returns the unchecked secondary even if the resource is unhealthy.
Users must see a static error page from S3 only when the ALB that serves the site is unhealthy. Which Route 53 setup fits?
Active-passive failover: a primary failover record for the ALB with a health check, and a secondary record (an alias to the S3 website endpoint) that Route 53 returns only when the primary is unhealthy.
A Route 53 failover design must work during a Regional outage. Why rely on health checks instead of editing DNS records during the event?
Health checks and DNS answers run on the globally distributed data plane.
Record changes use the control plane in us-east-1, which an outage could impair.
An EC2 application must be prevented from resolving known malicious domains through VPC Resolver. Which Route 53 feature fits when security groups cannot filter that resolver traffic?
Associate a Resolver DNS Firewall rule group with the VPC and block matching domains. Security groups and network ACLs cannot block the Amazon-provided resolver path.
A team adds only an approved-domain ALLOW list to Route 53 DNS Firewall and expects all other domains to be denied. What else is required?
A later rule blocking the remaining domains. Queries that match no rule resolve normally.
A central security account must reuse Route 53 DNS Firewall rules across application accounts. Which sharing and enforcement options apply?
Share rule groups with AWS RAM for recipient VPC associations. Use Firewall Manager policies to manage associations across the organization. Recipients cannot edit the shared group.
Does Route 53 DNS Firewall inspect HTTPS payloads or block an IP address obtained through another path?
No. It filters domain-name queries through VPC Resolver, not HTTPS payloads or resolved destination IP addresses.
A team needs the names and response codes of public DNS queries received by Route 53. What should it configure?
Enable public hosted-zone query logging to CloudWatch Logs with a log group in us-east-1 and permission for Route 53 to write to it.
Public Route 53 query logs show fewer lookups than application requests. Does that prove logging is broken?
No. Resolvers can answer from cached DNS data without sending another query to Route 53. Those lookups do not appear in public query logs.
Where does Route 53 send public hosted-zone query logs, and how are streams separated?
To CloudWatch Logs in us-east-1, with a stream for each Route 53 edge location answering for the zone.
A company shares Route 53 Resolver forwarding rules across accounts for workloads in multiple Regions. Can one Regional rule be associated with VPCs in every Region?
No. Resolver rules and their VPC associations are Regional. Share the relevant rules through AWS RAM and associate them with recipient VPCs.
A recipient needs a shared Route 53 Resolver forwarding rule for a VPC in a Region. What must it do after the rule is shared through AWS RAM?
Associate the shared rule with the VPC. Resolver rules and their VPC associations are Regional.
A multi-Region team uses Route 53 Resolver rules shared through AWS RAM. Can a recipient account edit a shared rule for its Regional VPCs?
No. Recipients can associate a shared rule with VPCs, but they cannot change or delete the rule.
Can a Route 53 private hosted zone alias to CloudFront and evaluate its target health?
No. Private zones cannot contain CloudFront aliases, and a CloudFront alias cannot enable Evaluate Target Health.
A Route 53 ALB alias has Evaluate Target Health enabled. One associated target group has no registered targets. Can it make the alias unhealthy?
Yes. An empty target group is unhealthy even if another target group has healthy targets.
A team needs application-aware DNS failover for API Gateway. Is Evaluate Target Health sufficient?
Use a Route 53 health check that monitors application access. AWS says inherited target health for API Gateway provides no operational benefit for this purpose.
A team wants to test Route 53 failover while its primary endpoint remains healthy. Which health-check setting can make it appear unhealthy?
Invert the healthy primary check. Disabling a non-inverted check makes it healthy and will not trigger this failover.
Both Route 53 primary and secondary failover records have failed their health checks. Which record is returned?
The primary record. Route 53 does not suppress both answers.
How does Route 53 aggregate endpoint health check results?
The endpoint is healthy when more than 18% of checkers report healthy. Each checker applies its interval and consecutive-check threshold.
What must a Route 53 HTTP or HTTPS health check receive to pass?
A TCP connection within four seconds and a 2xx or 3xx status within two more seconds. String checks need the body within another two seconds and the string within its first 5,120 bytes. HTTPS does not validate certificates.
How can the Route 53 health-check console notify an operator of an endpoint failure?
Create its default CloudWatch alarm and send notifications to an SNS topic in us-east-1. The default alarm waits for one minute of unhealthy status.
A Route 53 HTTPS check stays healthy after a certificate expires. Is certificate validation malfunctioning?
No. HTTPS health checks do not validate TLS certificates.
A Route 53 DNS Firewall ALERT rule matches a query before a later BLOCK rule. Does evaluation continue to the BLOCK rule?
No. ALERT permits the query, records an alert in configured Resolver logs, and stops inspection. Put rules in the intended priority order.
A protected VPC receives no response from Route 53 DNS Firewall. What is the default result, and how can availability be favored?
The default fail-closed mode returns SERVFAIL. Enable FirewallFailOpen to allow queries through when the firewall does not respond.
How do the Route 53 DNS Firewall NODATA, NXDOMAIN, and OVERRIDE block responses differ?
NODATA returns success without answer data. NXDOMAIN reports a nonexistent domain. OVERRIDE returns a custom CNAME with a default TTL of zero.
A Route 53 DNS Firewall rule lists example.com, but queries for its subdomains are still allowed. What list entry is missing?
A matching subdomain entry, such as *.example.com for all subdomains. The parent domain entry alone does not cover them.
A team wants to test a Route 53 DNS Firewall blocking policy without denying matching queries. What should it configure?
Use ALERT first and enable Resolver query logging for the VPC. Logs can go to CloudWatch Logs, S3, or Firehose.
What does Route 53 multivalue answer routing return when resources are healthy, and when all are unhealthy?
Up to eight healthy records, or all healthy records if there are eight or fewer. If all are unhealthy, it returns up to eight unhealthy records.
Every Route 53 weighted record has weight 0. Does the group stop answering?
No. All resources have equal selection probability.
A Route 53 latency alias points to a regional weighted group with no healthy endpoints. How can DNS choose another Region?
Enable Evaluate Target Health on the latency aliases and attach health checks to the non-alias weighted records. Route 53 can leave the failed regional branch.
What weight range does a Route 53 weighted record accept?
An integer from 0 through 255.
A healthy Route 53 weighted record has weight 0 while all records with nonzero weights fail their checks. Can it answer?
Yes. Route 53 considers zero-weight records when every nonzero-weight record is unhealthy.
A team calls SetAlarmState to test Route 53 failover, but the health check does not change. Why?
Route 53 reads the metric data stream, not the alarm state set by the API.
An operator changes the CloudWatch alarm threshold used by a Route 53 check. What additional update is required?
Update the Route 53 health check after changing the alarm settings.
Does the Route 53 authoritative DNS SLA guarantee its console and API availability?
No. The SLA is scoped to authoritative DNS hosted-zone availability and excludes API and console availability.
What does a team need to sign a public Route 53 hosted zone with DNSSEC?
A key-signing key backed by an asymmetric customer managed KMS key, then a DS record added at the parent zone. Private hosted zones cannot be signed.