Global TablesとRoute 53によるマルチサイト・アクティブ-アクティブ

マルチサイト・アクティブ-アクティブの定義

# Active-Active traffic split (normal operation):
# us-east-1: serving ~50% of users (North America)
# eu-west-1: serving ~50% of users (Europe)

# Active-Active traffic split (us-east-1 failure):
# us-east-1: 0% (health check failed)
# eu-west-1: 100% (ASG scales up automatically)

# RTO: near-zero (DNS TTL propagation only)
# RPO: near-zero (with DynamoDB Global Tables)

DynamoDB Global Tablesのアーキテクチャ

# Create DynamoDB table and add global regions
aws dynamodb create-table \
  --table-name UserSessions \
  --attribute-definitions AttributeName=userId,AttributeType=S \
  --key-schema AttributeName=userId,KeyType=HASH \
  --billing-mode PAY_PER_REQUEST \
  --region us-east-1

# Add replica regions for Global Table
aws dynamodb update-table \
  --table-name UserSessions \
  --replica-updates '[{"Create":{"RegionName":"eu-west-1"}},{"Create":{"RegionName":"ap-southeast-1"}}]' \
  --region us-east-1

書き込みアクティブ・読み取り対応のAurora Global Database

# Aurora Global Database read configuration
# Primary region (us-east-1): reads + writes
# Secondary region (eu-west-1): reads only
#   ~100ms replication lag, serves EU users low-latency reads

# Application reads from local Aurora endpoint
# Application writes to primary region Aurora endpoint

# Java connection string with region routing:
# readEndpoint=eu-west-1.cluster-ro-xxx.aurora.amazonaws.com
# writeEndpoint=us-east-1.cluster-xxx.aurora.amazonaws.com

Active-ActiveのためのRoute 53ルーティング

# Route 53 latency routing with health checks
aws route53 change-resource-record-sets \
  --hosted-zone-id ZXXX \
  --change-batch '{
    "Changes": [
      {
        "Action": "UPSERT",
        "ResourceRecordSet": {
          "Name": "api.example.com",
          "Type": "A",
          "Region": "us-east-1",
          "SetIdentifier": "us-east-1",
          "HealthCheckId": "hc-us-east-1",
          "AliasTarget": {"DNSName": "alb-us-east-1.amazonaws.com", "EvaluateTargetHealth": false}
        }
      }
    ]
  }'

トラフィック吸収のためのAuto Scaling

# ASG target tracking for request count
aws autoscaling put-scaling-policy \
  --auto-scaling-group-name app-asg-eu-west-1 \
  --policy-name scale-on-requests \
  --policy-type TargetTrackingScaling \
  --target-tracking-configuration '{
    "TargetValue": 1000,
    "PredefinedMetricSpecification": {
      "PredefinedMetricType": "ALBRequestCountPerTarget",
      "ResourceLabel": "app/my-alb/xxx/targetgroup/my-tg/yyy"
    },
    "ScaleInCooldown": 60,
    "ScaleOutCooldown": 30
  }'

Active-Activeにおけるセッション管理

# DynamoDB Global Table for session storage
# Session item structure:
{
  'sessionId': 'sess-abc123',
  'userId': 'usr-456',
  'data': {'cart': [...], 'preferences': {}},
  'expiresAt': 1750000000,
  'lastUpdatedRegion': 'us-east-1'
}

# Application reads from local region DynamoDB
# Writes replicate to all regions within ~1 second
# No sticky sessions needed on the ALB

書き込みの競合と解決

# Avoid conflicts with conditional writes
aws dynamodb update-item \
  --table-name UserProfiles \
  --key '{"userId":{"S":"usr-123"}}' \
  --update-expression 'SET profileVersion = profileVersion + :inc, username = :name' \
  --condition-expression 'profileVersion = :expectedVersion' \
  --expression-attribute-values '{
    ":inc":{"N":"1"},
    ":name":{"S":"newname"},
    ":expectedVersion":{"N":"5"}
  }'
# If another region already updated version, this fails gracefully

Active-ActiveにおけるS3レプリケーション

# Bidirectional S3 replication
# Bucket A (us-east-1) replicates to Bucket B (eu-west-1)
# Bucket B (eu-west-1) replicates to Bucket A (us-east-1)

# Enable S3 RTC for guaranteed replication time
aws s3api put-bucket-replication \
  --bucket us-east-1-uploads \
  --replication-configuration '{
    "Rules": [{
      "Status": "Enabled",
      "ReplicationTime": {"Status": "Enabled", "Time": {"Minutes": 15}},
      "Metrics": {"Status": "Enabled", "EventThreshold": {"Minutes": 15}},
      "Destination": {"Bucket": "arn:aws:s3:::eu-west-1-uploads"}
    }]
  }'

マルチリージョンオリジンでのCloudFront

# CloudFront origin group for multi-region failover
aws cloudfront create-distribution \
  --distribution-config '{
    "Origins": {
      "Quantity": 2,
      "Items": [
        {"Id": "us-east-1", "DomainName": "alb-us-east-1.amazonaws.com"},
        {"Id": "eu-west-1", "DomainName": "alb-eu-west-1.amazonaws.com"}
      ]
    },
    "OriginGroups": {
      "Items": [{
        "Id": "multi-region-group",
        "FailoverCriteria": {"StatusCodes": {"Items": [500,502,503,504]}},
        "Members": {"Items": [{"OriginId": "us-east-1"},{"OriginId": "eu-west-1"}]}
      }]
    }
  }'

Active-Activeのヘルス監視

# CloudWatch alarm for DynamoDB Global Table replication lag
aws cloudwatch put-metric-alarm \
  --alarm-name 'GlobalTable-ReplicationLag-eu-west-1' \
  --metric-name ReplicationLatency \
  --namespace AWS/DynamoDB \
  --dimensions Name=TableName,Value=UserSessions Name=ReceivingRegion,Value=eu-west-1 \
  --period 60 \
  --evaluation-periods 3 \
  --threshold 5000 \
  --comparison-operator GreaterThanThreshold \
  --alarm-actions arn:aws:sns:us-east-1:123:ops-alerts

Active-Activeを選ぶべきケース

# Active-Active justification checklist:
# [ ] Users in 2+ continents with latency SLAs
# [ ] RTO requirement < 5 minutes
# [ ] Revenue impact of downtime justifies 2x+ cost
# [ ] Data must remain within specific regions (regulations)
# [ ] Write throughput exceeds single-region capacity

# If fewer than 2-3 boxes checked:
# Consider Warm Standby instead (lower cost, adequate RTO)

クイックチェック

レッスンのまとめ