Skip to content

Aurora

Overview

  • Cluster Mode
  • A single primary instance + 0 or more replicas
    • primary instance - reads and writes
    • secondary instance - only reads
    • upto 15 replicas - can be anyone to be primary during failover

  • No local storage - uses cluster volume (SSD based)

    • faster provisioning & improved availability & performance
    • All SSD Based - high IOPs, low latency
    • Pay for only what is being used
    • Volumes are replicated synchronously
    • High water mark - billed for most used
    • No performance degradation during replication
    • Replicas can be added and removed without requiring storage provisioning

  • Endpoints

    • Cluster Endpoint - points to primary instance
    • Reader Endpoint - pooints to replicas (load-balance between multiple replicas)
    • Custom Endpoint - custom endpoint using db identifiers

Cost

  • No Free Tier Option
  • Does not support micro instances
  • Beyond RDS singleAZ (micro) Aurora offers better value
  • Compute - hourly charge, per second, 10 minutes minimum
  • Storage - GB-Month consumed, IO cost per request
  • 100% DB Size in backups are included

Restore, Clone and Backtrack

  • Backups in Aurora work in the same ways as RDS
  • Restores creates a new cluster
  • Backtrack can be used which allow in-place rewinds to a previous point in time
  • Fast clone make a new database MUCH faster than copying all the data - copy-on-write
    • just refrence the origin storage
    • store the difference

Aurora Serverless

  • Scalable => ACU - Aurora Capacity Units
    • MIN & MAX ACU is set for cluster
    • Scalable between MIN and MAX based on the load
  • Consumption billing per-second basis
  • Same reslience as Aurora (6 copies accross AZs)

Use Cases

  • Infrequently used applications
  • New applications
  • Variable workloads
  • Unpredictable workloads
  • Development and test databases
  • Multi-tenant applications

Aurora Multi-Master

  • Default Aurora mode is Single-Master
  • Failover takes time - replica promoted to R/W
  • In Multi-Master mode all instances are R/W
    • quoram of nodes to agree on the data to be written
  • Traffic is instantly switched to different master instance if one master fails