What's changed: Deepened SOA-C02 Chapter 2 to Associate depth (tables, scenarios, FAQ, traps; localized figures)
2.2Database Availability and Scaling
Understand RDS Multi-AZ (HA via automatic failover) vs. read replicas (read scaling) and when to use each.
Database availability and performance are handled separately. Learn them by role: HA = Multi-AZ, read scaling = read replicas.
2.2.1Multi-AZ and read replicas
- Multi-AZ: a synchronous standby in another AZ with automatic failover (HA); the standby is not readable.
- Read replicas: readable copies that offload reads (read scaling); asynchronous and not for failover.
A classic SOA trap is confusing HA/automatic failover = Multi-AZ (standby not readable) with offload read load = read replicas.
They’re easy to mix up, so learn them by role. Multi-AZ is for availability: a synchronously replicated standby in another AZ, with automatic failover by switching the DNS endpoint on primary failure. The standby is not usable for reads (standby only). Read replicas, by contrast, are for performance (read scaling): up to several asynchronously replicated, readable copies that offload read queries (cross-region too)—not a failover target. Combine both if you need each. For backups, automated backups plus transaction logs enable point-in-time recovery (PITR), while manual snapshots persist until you delete them. Aurora keeps six storage copies, promotes replicas fast, and exceeds RDS in both availability and performance. Patching and scaling during the maintenance window have less impact with Multi-AZ.
| Aspect | Multi-AZ | Read replica |
|---|---|---|
| Purpose | High availability | Read scaling |
| Replication | Synchronous | Asynchronous |
| Reads | No (standby) | Yes |
| Failover | Automatic | Not for it (manual promote) |
Scenario: the DB is a single point of failure and reads are heavy. Use Multi-AZ for automatic failover (availability) and add several read replicas to offload reads (performance). For DR, use cross-region read replicas or cross-region snapshot copies, and keep PITR to recover from mistakes.
Q. Automatic failover? Multi-AZ (standby not readable). Q. Reduce read load? Read replicas. Q. Recover mistakenly deleted data? Automated backups + PITR. Q. DR to another region? Cross-region replica/snapshot copy.
Watch the mix-ups: (1) A Multi-AZ standby is not readable—"read scaling = Multi-AZ" is wrong (it’s read replicas). (2) Read replicas are async and not a failover target (manual promotion is possible). (3) Disabling automated backups disables PITR.
If you need both HA and read scaling, combine Multi-AZ with read replicas. Aurora offers even faster replication and failover.
2.2.2Section summary
- Multi-AZ = HA/auto failover (not readable) / read replicas = read scaling
- Combine both if you need each
Sign in to track progress — Log in.
Quick check
(just a quick review)Q1. You want RDS to auto-fail over to a standby on primary failure for higher availability. What do you use?
Q2. Read queries are heavy and the primary is loaded. What offloads reads?
Q3. Which statement about a Multi-AZ standby is correct?