Amazon Aurora

Amazon Aurora is AWS's cloud-native relational database, MySQL- and PostgreSQL-compatible, delivered through Amazon RDS. It combines the familiarity of open-source engines with a re-architected storage layer that distributes data across three Availability Zones with six-way replication — giving higher throughput, faster recovery, and near-zero data loss.

Key Features:

• Distributed Storage: 10 GB segments replicated 6 ways across 3 AZs; storage auto-scales up to 128 TiB (Aurora MySQL/PostgreSQL) without downtime.
• Aurora Replicas: Up to 15 low-latency read replicas sharing the same storage volume; failover to a replica typically under 30 seconds.
• Aurora Serverless v2: On-demand auto-scaling capacity (measured in ACUs) that scales in seconds, ideal for variable or unpredictable workloads.
• Global Database: Single primary region with up to 5 read-only secondary regions; cross-region replication via dedicated infrastructure with sub-second lag and ~1-minute RTO on regional failure.
• Backtrack (MySQL): Rewind the database in place to a previous point in time (up to 72 hours) without restoring a backup.
• Parallel Query: Push query processing down into the distributed storage layer for large analytical scans.
• Zero-ETL to Redshift: Continuous replication of Aurora tables into Amazon Redshift for analytics without building pipelines.
• Aurora I/O-Optimized: Storage tier with no per-IO charges — flat predictable cost for I/O-heavy workloads.

Aurora vs. RDS (MySQL/Postgres):

• Performance: Aurora typically delivers 3-5x MySQL throughput and 2-3x PostgreSQL throughput on the same hardware.
• Availability: 6-way replication across 3 AZs vs. RDS's single-AZ primary + Multi-AZ standby.
• Cost: Aurora is ~20% more expensive than equivalent RDS, but the storage model often reduces total cost at scale.
• Choose RDS for plain MySQL/PostgreSQL/MariaDB/Oracle/SQL Server workloads; choose Aurora when you need Aurora-specific features (Serverless v2, Global Database, Backtrack, Parallel Query, Zero-ETL).

Common Use Cases:

• High-throughput OLTP workloads that outgrow vanilla RDS.
• SaaS applications needing multi-tenant scale with per-tenant read replicas.
• Global applications with sub-second cross-region read latency requirements.
• Variable workloads (dev/test, spiky traffic) on Aurora Serverless v2.
• Analytics on operational data via Zero-ETL to Redshift.

Service Limits & Quotas:

• Storage: 128 TiB max per cluster volume (auto-grows in 10 GB segments).
• Aurora Replicas per cluster: 15.
• Global Database secondary regions: up to 5.
• Backtrack window: up to 72 hours (Aurora MySQL only).
• Backup retention: 1-35 days for automated backups.
• Serverless v2 capacity: 0.5 to 256 ACUs per writer/reader (1 ACU ~ 2 GiB RAM + proportional CPU).
• Connections: engine-dependent; PostgreSQL default max_connections scales with instance size.

Pricing Model:

• Instance hours: per-second by class (db.r6g, db.r7g, db.t4g, etc.); Serverless v2 billed per ACU-hour.
• Storage (Standard): $0.10/GB-month + $0.20 per million IO requests.
• Storage (I/O-Optimized): $0.225/GB-month, no per-IO charge — wins above ~25% I/O-cost ratio.
• Backups: free up to cluster size; charged per GB-month beyond.
• Global Database: additional cross-region replication and storage charges.
• Common cost surprise: Standard storage on a heavy-write workload — IO costs can equal or exceed compute. Switch to I/O-Optimized once monthly IO charges exceed ~25% of compute.

Code Example — Create a Serverless v2 Aurora PostgreSQL Cluster:

aws rds create-db-cluster \
  --db-cluster-identifier prod-app \
  --engine aurora-postgresql \
  --engine-version 16.3 \
  --serverless-v2-scaling-configuration MinCapacity=0.5,MaxCapacity=16 \
  --master-username dbadmin \
  --manage-master-user-password \
  --vpc-security-group-ids sg-0abc123 \
  --db-subnet-group-name prod-private \
  --storage-encrypted \
  --kms-key-id alias/aurora \
  --backup-retention-period 14 \
  --deletion-protection

aws rds create-db-instance \
  --db-instance-identifier prod-app-writer \
  --db-cluster-identifier prod-app \
  --db-instance-class db.serverless \
  --engine aurora-postgresql

aws rds create-db-instance \
  --db-instance-identifier prod-app-reader \
  --db-cluster-identifier prod-app \
  --db-instance-class db.serverless \
  --engine aurora-postgresql
  

Connect via boto3 RDS Data API (Serverless):

import boto3

data = boto3.client("rds-data", region_name="us-west-2")
resp = data.execute_statement(
    resourceArn="arn:aws:rds:us-west-2:111122223333:cluster:prod-app",
    secretArn="arn:aws:secretsmanager:us-west-2:111122223333:secret:rds!cluster-xxx",
    database="orders",
    sql="SELECT id, total FROM orders WHERE created_at > :since",
    parameters=[{"name": "since", "value": {"stringValue": "2026-04-01"}}],
)
for row in resp["records"]:
    print(row)
  

Common Interview Questions: