What Is Amazon Route 53?

Amazon Route 53 is a highly available and scalable Domain Name System (DNS) web service that translates domain names into IP addresses and routes end users to applications.

What Problems Route 53 Solves:

• DNS reliability: Traditional DNS vulnerable to outages; Route 53 provides 100% uptime SLA
• Global traffic distribution: Manual traffic routing complex; Route 53 automates based on latency, geography, health
• Failover complexity: Detecting and routing around failures requires monitoring; Route 53 health checks automate failover
• Multi-region applications: Directing users to nearest region manually is impractical; Route 53 latency-based routing automates
• Traffic testing: A/B testing and blue/green deployments require traffic splitting; Route 53 weighted routing enables controlled rollouts

When to use Route 53:

• You need highly available DNS with 100% uptime SLA
• You require automated failover for multi-region applications
• You want to route users to the lowest-latency endpoint
• You need geolocation-based content delivery
• You want to implement blue/green deployments or A/B testing

Routing Policies

Route 53 offers seven routing policies for different traffic management scenarios.

Simple Routing

One-to-one mapping between domain and single resource.

How It Works:

• Returns single resource (IP address, load balancer, CloudFront distribution)
• No health checks
• If multiple values specified, returns all values in random order (client chooses)

Use Cases:

• Single web server
• Static websites
• Applications without redundancy

Example:

• example.com → 203.0.113.5

Weighted Routing

Distribute traffic across multiple resources in specified proportions.

How It Works:

• Assign weight to each record (0-255)
• Traffic percentage = Record weight / Sum of all weights
• Can associate health checks (skip unhealthy resources)

Use Cases:

• A/B testing: 90% production, 10% new version
• Blue/green deployments: Gradually shift traffic from blue to green
• Load distribution: Unequal capacity across regions (70% us-east-1, 30% eu-west-1)

Example:

• Record A (weight 7): 70% traffic → us-east-1 load balancer
• Record B (weight 3): 30% traffic → eu-west-1 load balancer

Cost: No additional charge beyond standard query fees.

Latency-Based Routing

Route users to lowest-latency AWS Region.

How It Works:

• Route 53 measures latency from user’s location to each AWS Region
• Returns resource in Region with lowest latency
• Based on actual latency measurements, not geographic proximity

Use Cases:

• Multi-region applications prioritizing performance
• Global user base with uneven distribution
• Applications where speed matters more than data residency

Example:

• User in Tokyo → Routes to ap-northeast-1 (lowest latency)
• User in London → Routes to eu-west-2 (lowest latency)
• User in New York → Routes to us-east-1 (lowest latency)

Performance: Typically reduces latency by 50-70% vs single-region deployment.

Failover Routing

Active-passive failover for high availability.

How It Works:

• Define primary and secondary resources
• Health check monitors primary resource
• If primary fails, Route 53 automatically returns secondary
• Failback when primary recovers

Use Cases:

• Disaster recovery: Primary region (us-east-1) fails → Secondary region (us-west-2)
• Maintenance windows: Direct traffic to secondary during primary maintenance
• Active-passive architecture: Database read replicas, standby servers

Example:

• Primary: us-east-1 load balancer (health check: HTTPS on /health)
• Secondary: us-west-2 load balancer (no health check)
• Primary fails → Traffic routes to us-west-2

Failover Time: 30-60 seconds (health check frequency + TTL)

Geolocation Routing

Route based on user’s geographic location.

How It Works:

• Route 53 identifies user’s location (continent, country, state)
• Returns resource mapped to that location
• Can define default location (catch-all)

Use Cases:

• Content localization: Serve region-specific content (language, currency, pricing)
• Data residency: Keep EU users’ data in EU regions (GDPR compliance)
• License restrictions: Block access from specific countries
• Load distribution: Regional load balancers

Example:

• Users in EU → eu-west-1 load balancer (GDPR compliance)
• Users in US → us-east-1 load balancer
• Users in Asia → ap-southeast-1 load balancer
• Default → us-east-1 load balancer

Granularity: Continent → Country → State (US only)

Geoproximity Routing (2024 Enhancement)

Route based on geographic location with bias adjustments.

How It Works:

• Routes to nearest resource by default
• Apply bias (+/-99) to expand or shrink geographic region
• Positive bias: Attract more traffic from farther away
• Negative bias: Reduce traffic from nearby areas

Use Cases:

• Cost optimization: Shift traffic to cheaper regions
• Capacity management: Reduce load on constrained resources
• Testing: Gradually expand new region’s coverage
• Data residency with flexibility: Prefer local resources but allow overflow

Example:

• us-east-1 (bias +50): Expanded coverage, attracts traffic from wider area
• eu-west-1 (bias 0): Standard coverage
• ap-southeast-1 (bias -25): Reduced coverage, only serves nearby users

2024 Update: Expanded availability from Traffic Flow to all DNS records (public and private hosted zones).

Multivalue Answer Routing

Return multiple IP addresses with health checks.

How It Works:

• Returns up to 8 healthy records randomly selected
• Each record has own health check
• Client chooses from returned values
• Unhealthy records excluded

Use Cases:

• Simple load distribution: Multiple web servers without load balancer
• Cost optimization: Avoid load balancer costs for small applications
• DNS-based availability: Client-side failover

Example:

• 10 web server IP addresses with health checks
• Route 53 returns 8 healthy IPs
• Client connects to one randomly

vs Weighted Routing: Multivalue is simpler (equal distribution, no weight configuration).

Health Checks

Health checks monitor resource availability and enable automated failover.

Types of Health Checks

1. Endpoint Health Checks:

• Monitor HTTP/HTTPS/TCP endpoints
• Specified by IP address or domain name
• Check interval: 10 seconds (fast) or 30 seconds (standard)
• Failure threshold: 3 consecutive failures = unhealthy

2. Calculated Health Checks:

• Monitor status of other health checks
• Use Boolean logic (AND, OR, NOT)
• Example: Require 2 of 3 child health checks healthy

3. CloudWatch Alarm Health Checks:

• Monitor CloudWatch alarms
• Use any CloudWatch metric
• Example: Lambda error rate, DynamoDB throttles

Health Check Configuration

Endpoint Health Check Parameters:

• Protocol: HTTP, HTTPS, TCP
• Port: Default 80 (HTTP), 443 (HTTPS), or custom
• Path: /health or custom health check endpoint
• Interval: 30 seconds (standard, $0.50/month) or 10 seconds (fast, $1/month)
• Failure threshold: Default 3 (1-10 allowed)
• String matching: Optional HTTP response body check

Health Check Regions:

• Route 53 checks from multiple global locations (15+ regions)
• Majority consensus determines health status
• Prevents false positives from single location issues

Failover Scenarios

Active-Passive Failover:

• Primary resource with health check
• Secondary resource (no health check, always considered healthy)
• Primary unhealthy → Traffic routes to secondary

Active-Active Failover:

• Multiple resources, each with health check
• Traffic distributed across healthy resources
• Unhealthy resources automatically excluded

Combination Failover:

• Mix routing policies (latency + failover, weighted + health checks)
• Complex multi-region architectures
• Example: Latency-based routing with failover per region

Health Check Pricing (2024)

• Standard health checks (30s interval): $0.50/month per health check
• Fast health checks (10s interval): $1.00/month per health check
• Calculated health checks: $0.50/month per health check
• CloudWatch alarm health checks: $0.50/month per health check

Example Cost:

• 10 endpoints × $0.50 = $5/month
• High availability across 3 regions (6 endpoints + 3 calculated) = $4.50/month

DNS Fundamentals

Record Types

A Record: IPv4 address

• example.com → 203.0.113.5

AAAA Record: IPv6 address

• example.com → 2001:0db8:85a3::8a2e:0370:7334

CNAME Record: Alias to another domain

• www.example.com → example.com
• Cannot be used for zone apex (example.com)

Alias Record (AWS-specific):

• Points to AWS resources (CloudFront, ALB, S3, API Gateway)
• Can be used for zone apex
• Free queries (no charge for Alias record queries to AWS resources)

MX Record: Mail exchange

• Priority + mail server
• 10 mail.example.com

TXT Record: Text information

• SPF, DKIM, domain verification

TTL (Time to Live)

Controls how long DNS resolvers cache the record.

Short TTL (60-300 seconds):

• Faster propagation of changes
• Higher query costs (more queries to Route 53)
• Use for: Frequently changing IPs, testing, deployments

Long TTL (3600-86400 seconds):

• Lower query costs (fewer queries)
• Slower propagation of changes
• Use for: Stable resources, cost optimization

Alias Records: TTL managed by Route 53 (cannot be changed).

Best Practice: Start with short TTL (60s) during testing, increase to long TTL (3600s+) for production stability.

Hosted Zones

Container for DNS records for a domain.

Public Hosted Zones

DNS records accessible from the internet.

Pricing:

• $0.50/month per hosted zone
• $0.40 per million queries (first 1 billion/month)
• $0.20 per million queries (after 1 billion/month)

Use Cases:

• Public-facing websites
• APIs accessible from internet
• Email servers

Private Hosted Zones

DNS records accessible only within VPCs.

Pricing:

• $0.50/month per hosted zone
• Queries are FREE (no per-query charges)

Use Cases:

• Internal services (databases, microservices)
• Private APIs
• Service discovery within VPC

Configuration:

• Associate with one or more VPCs
• Can span multiple AWS accounts (VPC sharing)

Best Practice: Use private hosted zones for internal services to avoid exposing internal DNS and save on query costs.

Cost Optimization

Strategies to Reduce Route 53 Costs

1. Use Alias Records for AWS Resources:

• Queries to Alias records pointing to AWS resources are FREE
• Standard A/AAAA records: $0.40 per million queries
• Savings: 100% on queries to CloudFront, ALB, S3, API Gateway

2. Increase TTL Values:

• Higher TTL = fewer queries = lower costs
• 60s TTL: ~1.4 billion queries/month for 1,000 req/s traffic
• 3600s TTL: ~24 million queries/month for same traffic
• Savings: 98% query reduction

3. Use Private Hosted Zones for Internal Services:

• Public hosted zone: $0.50/month + $0.40 per million queries
• Private hosted zone: $0.50/month + FREE queries
• Internal services with 1 billion queries: $400/month savings

4. Consolidate Hosted Zones:

• Multiple subdomains: Use single hosted zone with multiple records
• Example: api.example.com, www.example.com, cdn.example.com → One hosted zone
• Savings: $0.50/month per consolidated domain

5. Delete Unused Hosted Zones:

• Audit monthly billing for unused zones
• Each unused zone: $0.50/month wasted

6. Share Resolver Endpoints Across Accounts:

• Resolver endpoint: $0.125/hour per ENI = $91/month per endpoint
• Share single endpoint across multiple VPCs/accounts (same region)
• Savings: $91/month per avoided endpoint

Cost Example

Scenario: Website with 100 million requests/month, CloudFront + ALB

Before Optimization:

• Hosted zone: $0.50
• A record to CloudFront: 100M queries × $0.40/M = $40
• A record to ALB: 20M queries × $0.40/M = $8
• Total: $48.50/month

After Optimization:

• Hosted zone: $0.50
• Alias to CloudFront: FREE
• Alias to ALB: FREE
• Increased TTL (300s → 3600s): 80% fewer queries
• Total: $0.50/month

Savings: $48/month (99%)

Common Pitfalls

Pitfall	Impact	Solution
1. Using A records instead of Alias for AWS resources	$40/100M queries wasted	Use Alias records (free queries to AWS resources)
2. Low TTL on stable resources	10-60x higher query costs	Increase TTL to 3600s+ for production (98% cost reduction)
3. No health checks for failover	Manual failover, downtime during outages	Configure health checks ($0.50/month, automated failover)
4. Public hosted zone for internal services	$0.40/M queries + security risk	Use private hosted zones (free queries, VPC-only access)
5. No default geolocation record	Users in unmapped locations get NXDOMAIN	Always define default location (catch-all)
6. Single-region deployment	Users far from region experience high latency	Use latency-based routing across multiple regions
7. Not testing health checks	False positives/negatives, improper failover	Test health checks with intentional failures
8. Unused hosted zones	$0.50/month wasted per zone	Audit and delete unused zones monthly
9. Equal weighted routing for unequal capacity	Overloading smaller instances	Adjust weights based on capacity (70/30 vs 50/50)
10. No bias in geoproximity routing	Traffic distribution doesn’t match needs	Use bias to shift traffic to preferred regions
11. Missing secondary for failover	Failover incomplete (no fallback)	Always define secondary resource
12. Health check interval too slow	30s+ detection delay	Use fast health checks (10s) for critical apps ($1/month)
13. Not using private hosted zones	Exposing internal DNS, higher costs	Create private hosted zones for internal services
14. Complex routing without testing	Unexpected traffic patterns	Test routing policies in dev/staging first
15. No CloudWatch alarms for health checks	Unnoticed health check failures	Create alarms for HealthCheckStatus metric

Cost Impact Examples:

• Pitfall #1 (A vs Alias): 100M queries = $40/month wasted
• Pitfall #2 (Low TTL): 60s → 3600s TTL = 98% query reduction
• Pitfall #4 (Public vs Private): 1B internal queries = $400/month wasted

Integration Patterns

Route 53 + CloudFront

Global Content Delivery:

• Route 53 Alias record → CloudFront distribution (free queries)
• Latency-based routing to multiple CloudFront distributions (multi-region)
• Geolocation routing for compliance (serve EU content from EU distribution)

Route 53 + Application Load Balancer

Multi-Region Load Balancing:

• Latency-based routing → ALB in each region
• Health checks on ALB targets
• Automatic failover if region becomes unhealthy

Blue/Green Deployments:

• Weighted routing: 90% blue ALB, 10% green ALB
• Gradually shift weights: 70/30, 50/50, 30/70, 0/100
• Rollback: Shift weight back to blue

Route 53 + API Gateway

API Traffic Management:

• Alias record → API Gateway (free queries)
• Weighted routing for API version testing (v1: 95%, v2: 5%)
• Geolocation routing for regional APIs (GDPR compliance)

Route 53 + Multi-Region Databases

Read Replica Routing:

• Latency-based routing to RDS read replicas in each region
• Health checks on each replica
• Lowest-latency reads for global applications

Failover to Secondary Region:

• Primary record → us-east-1 Aurora cluster (health check)
• Secondary record → us-west-2 Aurora cluster
• Automatic failover if primary cluster fails

Key Takeaways

Routing Policies:

• Simple: Single resource, no health checks
• Weighted: A/B testing, blue/green (10% new version, 90% old)
• Latency-Based: Route to lowest-latency region (50-70% latency reduction)
• Failover: Active-passive disaster recovery (30-60s failover)
• Geolocation: Content localization, data residency (GDPR)
• Geoproximity: Location-based with bias (2024: available for all records)
• Multivalue: Simple load distribution (up to 8 healthy IPs)

Health Checks:

• Monitor HTTP/HTTPS/TCP endpoints ($0.50/month standard, $1/month fast)
• Enable automated failover (no manual intervention)
• Check from 15+ global locations (majority consensus)
• Calculated health checks combine multiple checks (Boolean logic)

Cost Optimization:

• Use Alias records for AWS resources (100% query cost savings)
• Increase TTL for stable resources (98% query reduction: 60s → 3600s)
• Private hosted zones for internal services (free queries vs $0.40/M)
• Consolidate hosted zones ($0.50/month per zone)

High Availability:

• Failover routing with health checks (30-60s automated failover)
• Multi-region latency-based routing (50-70% latency improvement)
• Active-active with weighted routing + health checks

100% Uptime SLA:

• Route 53 is the only AWS service with 100% availability SLA
• Globally distributed infrastructure (multiple edge locations)
• No single point of failure

Pricing:

• Hosted zones: $0.50/month
• Queries: $0.40 per million (first 1B), $0.20 per million (after 1B)
• Alias queries to AWS resources: FREE
• Health checks: $0.50-$1/month each