AWS Knowledge Hub

English

Tiếng Việt

English

Tiếng Việt

Appearance

Amazon Aurora ​

Amazon Aurora is a MySQL and PostgreSQL-compatible relational database engine that combines the performance and availability of high-end commercial databases with the simplicity and cost-effectiveness of open-source databases.

Overview ​

Aurora is built for the cloud, with a distributed, fault-tolerant, self-healing storage system that auto-scales up to 128 TB per database instance. It provides up to 15 low-latency read replicas, point-in-time recovery, continuous backup to S3, and replication across three Availability Zones.

Key Features ​

  • High Performance: Up to 5x better performance than MySQL, 3x better than PostgreSQL
  • High Availability: 99.99% availability, automatic failover
  • Auto Scaling Storage: Automatically grows from 10 GB to 128 TB
  • Fast Replication: Low-latency read replicas (typically < 100ms)
  • Backup and Recovery: Continuous backup to S3, point-in-time recovery
  • Multi-Master: Aurora Multi-Master for read/write scaling
  • Serverless: Aurora Serverless for variable workloads
  • Global Database: Cross-region replication with < 1 second replication lag

Database Engines ​

Aurora MySQL ​

  • MySQL 5.7 and 8.0 compatible
  • Up to 5x better performance than MySQL
  • Compatible with MySQL applications and tools

Aurora PostgreSQL ​

  • PostgreSQL 11, 12, 13, 14, 15 compatible
  • Up to 3x better performance than PostgreSQL
  • Compatible with PostgreSQL applications and tools

Architecture ​

Storage Layer ​

  • Distributed, self-healing storage across 3 AZs
  • 6 copies of data across 3 AZs
  • Automatic repair and recovery
  • No I/O freezing during backup

Compute Layer ​

  • DB instances (primary and replicas)
  • Independent scaling of compute and storage
  • Fast failover (typically < 30 seconds)

Deployment Options ​

Single-Master ​

  • One primary instance, multiple read replicas
  • Read replicas can be promoted to primary
  • Suitable for most workloads

Multi-Master ​

  • Multiple read/write instances
  • Automatic conflict detection and resolution
  • Suitable for applications with high write traffic
  • All instances can accept writes

Serverless ​

  • Automatically starts, scales, and shuts down
  • Pay per ACU (Aurora Capacity Unit) second
  • Suitable for variable or unpredictable workloads
  • No capacity planning required

Read Replicas ​

  • Up to 15 read replicas per cluster
  • Low-latency replication (typically < 100ms)
  • Can be in different regions
  • Automatic failover support
  • Can be promoted to standalone cluster

Global Database ​

  • Cross-region replication
  • Up to 16 secondary regions
  • < 1 second replication lag
  • Automatic failover to secondary region
  • Useful for disaster recovery and low latency

Backup and Recovery ​

Automated Backups ​

  • Continuous backup to S3
  • Point-in-time recovery to any second
  • Retention: 1-35 days
  • No performance impact

Manual Snapshots ​

  • User-initiated backups
  • Retained until explicitly deleted
  • Can be shared across accounts
  • Can be copied to other regions

Backtrack ​

  • Rewind database to any point in time
  • No need to restore from backup
  • Available for up to 72 hours
  • Useful for testing and recovery

Performance Features ​

Performance Insights ​

  • Real-time database performance monitoring
  • Identify bottlenecks and slow queries
  • Retention up to 24 months

Query Plan Management ​

  • Pin optimal query plans
  • Prevent plan regressions
  • Automatic plan evolution

Parallel Query ​

  • Parallel execution of analytical queries
  • Faster complex queries
  • Automatic parallelization

Security ​

  • Encryption at Rest: AWS KMS integration
  • Encryption in Transit: SSL/TLS connections
  • Network Isolation: VPC deployment
  • IAM Database Authentication: IAM users for database access
  • Audit Logging: Comprehensive audit logs

Use Cases ​

  • Enterprise Applications: High-performance database needs
  • Web Applications: Scalable backend databases
  • SaaS Applications: Multi-tenant databases
  • Analytics: Fast analytical queries with parallel query
  • Global Applications: Low-latency access with Global Database

Best Practices ​

  • Use Multi-AZ deployments for high availability
  • Enable automated backups with appropriate retention
  • Use read replicas for read scaling
  • Monitor performance with Performance Insights
  • Use Aurora Serverless for variable workloads
  • Implement proper security groups and encryption
  • Use Global Database for multi-region applications
  • Right-size instances based on workload
  • Enable backtrack for critical databases
  • Use parallel query for analytical workloads

Cost Optimization ​

  • Use Reserved Instances for predictable workloads
  • Use Aurora Serverless for variable workloads
  • Right-size instances to match actual usage
  • Use read replicas instead of larger instances
  • Delete unused snapshots
  • Monitor and optimize storage usage
  • Use Global Database only when needed

Comparison with RDS ​

Advantages ​

  • Better performance (up to 5x MySQL, 3x PostgreSQL)
  • Auto-scaling storage (10 GB to 128 TB)
  • Faster replication (< 100ms vs seconds)
  • More read replicas (15 vs 5)
  • Backtrack feature
  • Global Database for multi-region

When to Choose RDS ​

  • Need specific MySQL/PostgreSQL versions not in Aurora
  • Cost optimization for smaller workloads
  • Don't need Aurora-specific features