Aurora automatically adds or removes Read Replicas based on how busy your database is.
Client sends many requests
↓
Reader Endpoint notices high CPU usage
↓
Aurora automatically adds more replicas
↓
Reader Endpoint extended (now has more replicas)
↓
Load distributed across all replicas
↓
When traffic decreases, Aurora removes extra replicas
Monitoring:
Scaling Up:
Scaling Down:
Load Balancing: