Amazon RDS Instance
Does ZopNight manage Amazon RDS Instance?
An RDS instance bills instance-hours by DB instance class plus GB-months for provisioned storage, and Multi-AZ doubles the instance meter. ZopNight stops non-production databases on schedule via rds:StopDBInstance. Storage keeps billing while stopped, and AWS restarts any stopped instance automatically after 7 days.
Rules that fire on Amazon RDS Instance
At a glance
| Field | Value |
|---|---|
| Stop verb | rds:StopDBInstance |
| Scheduling notes | instance stopped via rds:StopDBInstance and started via rds:StartDBInstance on schedule. |
Amazon Relational Database Service (RDS) runs managed MySQL, PostgreSQL, MariaDB, Oracle, and SQL Server instances. Database instances bill hourly for compute plus storage whether queried or not, so non-production databases are among the highest-value scheduling targets.
Two meters, and Multi-AZ doubles one
An RDS instance bills instance-hours at a rate set by the DB instance class, plus GB-months for provisioned storage, plus any provisioned IOPS or throughput. Enabling Multi-AZ runs a synchronous standby, which doubles the instance-hour meter: correct for production, and a common accidental default on databases that never needed failover.
The 7-day stop limit
Stopping an instance halts the instance-hour meter, but two catches apply. Storage, backups, and snapshots keep billing while the database is stopped. And AWS automatically restarts any instance that has been stopped for 7 consecutive days. A stopped database is a temporary state, not a parked one. A scheduler that stops nightly and starts each morning never hits the limit; a one-off manual stop quietly comes back a week later.
Overspend patterns
The recurring ones: Multi-AZ and point-in-time recovery left enabled on dev and test databases; gp2 storage paying a higher rate than gp3 for the same capacity; backup retention set far beyond what non-production data justifies; and instances idling at zero connections for weeks because the application they served was retired.
Discovery, metrics, and scheduling
ZopNight finds instances every 6 hours through Resource Explorer 2 and the RDS API, keeps hourly CloudWatch metrics (CPU, connections, IOPS) with a 90-day lookback, and attributes cost per instance from Cost Explorer or CUR 2.0. Off-hours schedules stop with rds:StopDBInstance and start with rds:StartDBInstance; recommendations cover idle databases, rightsizing, and Graviton instance classes.
Console location
RDS console, then Databases. Each database’s Configuration tab shows the instance class, Multi-AZ setting, and storage type, the three fields that explain most surprising RDS bills. The list is per-region, like EC2’s.