Aurora

Introduction

  • A MySQL and PostgreSQL compatible relational database on AWS

  • 5x faster than standard MySQL databases and 3x faster than standard PostgreSQL databases.

  • Provides the security, availability, and reliability of commercial-grade databases at 1/10 of the cost.

Feature

  • Provision Type:

    • Self-Provisioned

      • Starts with 10 GB, scales in 10 GB increments to 64TB automatically

      • Can scale up to 32 CPUs and 244GB of memory

    • Serverless

      • Automated database instantiation and auto-scaling based on actual usage.

  • 2 copies of data in each AZ with minimum of 3 AZs (can choose AZ, and have 6 copies at least)

    • 4 copies out of 6 needed for writes

    • 3 copies out of 6 neede for reads

  • Read / Write

    • With a write endpoint to write into Master

    • With a read endpoint to read from all Read Replicas (Scalability):

      • Aurora Replica

        • up to 15 Read Replicas

      • MySQL Replica

        • up to 5 Read Replicas

  • Cross Region Read Replicas (RR): entire DB is copied (not select tables)

  • Import / export data directly from / to S3: efficient use of resources

  • Backtracking for rewinding the database status (useful for doing Tests)

    • Point-in-time restoration from snapshots creates a new database. (needs more time)

  • Global Aurora (cross region)

    • Aurora Cross Region Read Replicas

      • Useful for DR

      • Simple to put in place

    • Aurora Global Database (recommended)

      • One Primary Region (read / write)

      • Up to 5 secondary (read only) regions, replication lag is lesser than 1 second

      • Up to 16 Read Replicas per secondary region

      • Helpful for decreasing latency

      • Promoting another region (for DR) has an RTO within 1 minute

  • Aurora Multi-Master

    • In case you need immediate failover for write node (HA)

    • Every Aurora node can do read / write (need to promote a RR regularly)

    • Limits:

      • The maximum number of nodes: 2.

      • Can only be in one region.

      • May increase latency and deal with conflict with an algorithm.

      • Not support integration with other AWS services.

      • Etc.

  • Loss Handling

    • Loss of up to 2 copies of data without affecting write availability

    • Loss of up to 3 copies of data without affecting read availability

    • Self-healing (scanning errors and repairing automatically)

PreviousStorageNextDynamoDB

Last updated