Relational Database Service (RDS)

Icon

RDS Icon

  • RDS dont allow access to underlying EC2 instances. So one cannot SSH into the machine running the RDS, except when using RDS custom service.

Features

Storage Auto Scaling

  • This feature helps in increasing the storage on your DB instance dynamically.

  • When this feature is enabled RDS detects when the database is running out of space (based on threshold value: Maximum Storage Threshold), will automatically scale the storage.

  • No manual intervention is needed for scaling when this feature is enabled.

  • Maximum Storage Threshold parameter can be configured based on,

    • Percentage of space left.

    • Low storage last for a specific duration.

    • Time passed since last modification.

  • Useful for applications with unpredictable workload.

  • This feature is available for all RDS database engines.

Read Replicas

  • When an application is burdened by read and write operation on single node of database then, it is possible that single node may go down becoming single point of failure.

  • To avoid this, a common strategy is to horizontally scale the database instance to handle reads and writes to separate instances.

  • RDS allows upto 15 Read Replicas to scale the read operation.

  • These Read Replicas can be on same AZ, cross AZ or cross-region.

  • Replication done is ASYNC, so reads are eventually consistent.

  • Replicas can be promoted to their own database when needed, which may allow them to take writes.

  • If RDS Read Replication is within same region is free i.e different AZ, cross-region is chargeable.

  • More can be read here.

Multi AZ

  • This feature is used for disaster recovery and not for scaling.

  • With this feature, one get a standby instance.

  • One DNS name is associated with both master and stand by instance.

  • This increases availability of the service and provides failover in case of loss of AZ in which master instance resides.

  • Read Replicas can be setup as Multi AZ for Disaster Recovery (DR).

  • The replication between standby and master instance is synchronous.

  • Conversion of Single-AZ database instance to Multi-AZ can be done with zero downtime. Internally, it is basically

    • Taking a snapshot of the DB and restore it in new AZ.

    • Establish synchronization between two instances.

Last updated