RDS 扩展与性能调优

As your application grows, a single RDS instance may struggle to keep up with traffic. Scaling ensures your database stays responsive under load.

There are two broad strategies:

• Vertical scaling — make the instance bigger (more CPU/RAM)
• Horizontal scaling — spread reads across multiple instances

Vertical scaling means changing the DB instance class, for example moving from db.t3.medium to db.m6g.large.

RDS applies the change during a maintenance window or immediately, usually with a short restart.

RDS can automatically grow your storage when free space runs low. Enable Storage Autoscaling and set a maximum threshold so you never run out of disk unexpectedly.

• Set a max allocated storage cap
• RDS adds storage in increments as needed

For read-heavy workloads, create read replicas. Replicas serve SELECT queries while the primary handles writes.

Point reporting dashboards and analytics at replicas to take load off the primary.

Opening too many database connections wastes memory and can exhaust limits. Amazon RDS Proxy pools and reuses connections, which is especially helpful for serverless apps and Lambda.

• Reduces connection overhead
• Improves failover times

Enhanced Monitoring gives OS-level metrics (CPU, memory, disk I/O) at up to 1-second granularity. Combine it with CloudWatch to spot bottlenecks before they cause outages.

Performance Insights visualizes database load and shows you the top SQL statements, waits, and users driving load.

Use it to identify slow queries that need indexing or rewriting.

The single biggest performance win is often a good index. Indexes let the database find rows without scanning the whole table.

Add indexes on columns used in WHERE, JOIN, and ORDER BY clauses.

RDS exposes engine settings through DB parameter groups. Tune values like max_connections or buffer pool size to match your workload.

Create a custom parameter group rather than editing the default.

Not every query needs to hit the database. Put a cache like Amazon ElastiCache (Redis/Memcached) in front of RDS for frequently read, rarely changing data.

• Lower latency for hot data
• Fewer queries reaching RDS

Before scaling, follow this order:

• Profile with Performance Insights
• Add or fix indexes
• Offload reads to replicas
• Add caching
• Only then scale the instance up

Throwing a bigger instance at a missing index is expensive and rarely the real fix.

Test your scaling knowledge.

You learned to scale RDS effectively:

• Vertical scaling changes the instance class
• Read replicas scale reads horizontally
• RDS Proxy pools connections
• Performance Insights and indexing tune queries

Always profile first, fix indexes, offload and cache before paying for a bigger instance.