RDS

Managed relational databases — AWS runs the OS, patching, backups, and failover so you manage schema and queries, not servers.

• Engines: PostgreSQL, MySQL, MariaDB, Oracle, SQL Server, and Amazon Aurora (see Aurora).
• Provisioned on EC2 under the hood; you pick instance class + storage (gp3/io1/io2).
• Not serverless (except Aurora Serverless) — you choose and pay for the instance.

High Availability vs Read Scaling

These solve different problems — don’t confuse them.

	Multi-AZ	Read Replicas
Purpose	High availability / failover	Scale reads
Replication	Synchronous	Asynchronous
Readable?	No (standby is passive)	Yes
Failover	Automatic (~60–120s, DNS swap)	Manual promote
Span	2 or 3 AZs	Same region or cross-region

• Read replicas: up to 5 per source (15 for Aurora). Promote a replica to a standalone DB for DR or migration.
• Multi-AZ DB cluster (newer) gives 2 readable standbys with faster failover.

Backups & Recovery

• Automated backups — daily snapshot + transaction logs, retention up to 35 days, enables point-in-time recovery (PITR).
• Manual snapshots — kept until you delete them; can be shared/copied across accounts and regions.
• Restoring always creates a new instance.

Storage, Security, Performance

• Storage autoscaling raises allocated storage automatically under pressure.
• Encryption at rest via KMS (set at creation); TLS in transit.
• RDS Proxy — managed connection pooling; essential in front of Lambda to avoid connection exhaustion.
• Parameter groups (engine config) and option groups (engine features).
• Performance Insights — visualize DB load and top SQL.

Snippets

# Force a Multi-AZ / cluster failover to a reader
aws rds failover-db-cluster \
  --db-cluster-identifier db-cluster-name \
  --target-db-instance-identifier db-instance-reader-name
 
# Create a manual snapshot
aws rds create-db-snapshot \
  --db-instance-identifier my-db \
  --db-snapshot-identifier my-db-$(date +%Y%m%d)
 
# List instances and their endpoints
aws rds describe-db-instances \
  --query 'DBInstances[].{id:DBInstanceIdentifier,ep:Endpoint.Address,az:AvailabilityZone}' \
  --output table

When to leave RDS

For serverless/variable workloads or >64 TiB and global reads, reach for Aurora (Serverless v2, Global Database). For key-value access patterns at scale, DynamoDB.