How to Resize AWS RDS Instances in Production Without Causing Downtime
Resizing an Amazon RDS instance using the default in-place modification triggers 5 to 15 minutes of downtime as AWS reboots and migrates the database to a new host. To avoid this, engineers can use one of three zero-downtime strategies: RDS Blue/Green Deployments, read-replica promotion, or a Multi-AZ failover to a pre-resized standby. Blue/Green Deployments, now supported for MySQL, MariaDB, PostgreSQL, and SQL Server, is considered the preferred approach in 2026, with cutovers typically completing in under 60 seconds. Read-replica promotion suits older engine versions or cross-region scenarios, while the Multi-AZ failover method is best reserved for cases where neither other option is feasible. To reduce the need for emergency resizing altogether, AWS recommends monitoring Performance Insights weekly, using Graviton-based instance classes for better price-performance, and setting CloudWatch alarms to detect resource drift early.
This is an AI-generated summary. ShortSingh links to the original source for the complete article.

Discussion (0)
Log in to join the discussion and vote.
Log in