Amazon RDS & Multi-AZ

RDS runs six relational engines: PostgreSQL, MySQL, MariaDB, Oracle, Microsoft SQL Server, and Amazon Aurora (a cloud-native, MySQL- and PostgreSQL-compatible engine). The first five are sometimes called the community or commercial engines, while Aurora is AWS-built and offers higher performance and its own storage layer. Oracle and SQL Server are licensed engines, so you choose either License Included pricing or Bring Your Own License (BYOL). For the exam, recognize that a request to migrate an existing PostgreSQL, MySQL, MariaDB, Oracle, or SQL Server database with minimal changes maps directly to RDS.

A Multi-AZ deployment provisions a primary instance and a standby in a different Availability Zone (an isolated location within a region). Data is replicated synchronously, meaning a write must be committed on both the primary and the standby before it is acknowledged, which gives strong durability with no data loss on failover. The standby does not serve read or write traffic. If the primary fails (or during patching, instance changes, or an AZ disruption), RDS performs an automatic failover by repointing the database DNS endpoint to the standby, usually within one to two minutes. Because clients connect through that endpoint, application connection strings do not change.

A read replica is a read-only copy of the source database that uses asynchronous replication (changes are streamed after they commit on the source, so a replica can lag slightly behind). Read replicas offload read-heavy traffic such as reporting and analytics queries, and you can create up to 15 per source for the major open source engines. Each replica has its own endpoint, so the application must direct read queries to it. Replicas can be created in the same region or in a different region (cross-region), which also supports disaster recovery and low-latency reads near distant users. A read replica can be promoted to a standalone writable database, which is useful for migrations or regional failover. Read replicas are about scaling and offloading reads, while Multi-AZ is about automatic failover.

RDS Storage Auto Scaling lets the volume grow automatically when free space runs low, so you avoid out-of-space outages without overprovisioning. You enable it and set a maximum storage threshold (an upper cap RDS will not exceed). RDS expands storage when free space falls below about 10 percent of allocated capacity for at least five minutes, and it applies the increase without downtime. It works on the General Purpose SSD (gp2 and gp3) and Provisioned IOPS SSD (io1 and io2) storage types. To avoid runaway growth, RDS limits how often it scales (typically no more than once every six hours). Storage can grow automatically, while shrinking allocated storage is not supported.

A parameter group is a container of engine configuration values (for example max_connections or work_mem) that you attach to an instance to tune behavior without editing files on a server you do not control. Parameters are either dynamic (apply immediately) or static (require a database reboot to take effect). The default parameter group cannot be edited, so you create a custom one to make changes. An option group is related but separate: it enables optional engine features such as Oracle Transparent Data Encryption or SQL Server features. For the exam, remember that parameter groups change runtime settings and that static parameters need a reboot.