Scheduling Amazon Redshift Cluster
redshift:PauseClusterCan ZopNight schedule Amazon Redshift Cluster?
Redshift pauses via redshift:PauseCluster, halting the per-hour node charge, and unlike the stopped RDS engines a paused cluster stays paused indefinitely, with no 7-day forced restart. RA3 managed storage bills through the pause, resume takes minutes with the endpoint unchanged, and reserved nodes bill regardless of state.
How the stop works
| Field | Value |
|---|---|
| Behaviour | cluster paused via redshift:PauseCluster on stop and resumed via redshift:ResumeCluster on start. |
Example schedules
- 0 8 * * 1-5 — Business Hours Start: Start at 8:00 AM on weekdays
- 0 18 * * 1-5 — Business Hours Stop: Stop at 6:00 PM on weekdays
- 0 22 * * * — Night Shutdown: Stop at 10:00 PM every day
- 0 6 * * 1-5 — Morning Startup: Start at 6:00 AM on weekdays
- 0 20 * * 5 — Weekend Shutdown: Stop at 8:00 PM on Friday
- 0 7 * * 1 — Weekend Startup: Start at 7:00 AM on Monday
The node meter stops, indefinitely
redshift:PauseCluster ends on-demand compute billing for every node in the cluster, and,
unusually among AWS data services, there is no automatic restart deadline. A paused Redshift
cluster stays paused for weeks if that is what the calendar says. Weekend-only analytics
clusters and end-of-quarter reporting warehouses can sleep for the whole gap without the
seven-day babysitting the database engines require.
What a pause preserves
The cluster’s endpoint, parameter groups, users and data all survive. redshift:ResumeCluster
brings back the same cluster at the same address, so BI tools and ETL configs need no changes.
Snapshots continue on their own cadence and continue billing; scheduled maintenance that falls
inside a pause is deferred, so long pauses can accumulate pending maintenance for the next
resume window.
Managed storage bills through it
On RA3 node types, storage is metered separately as managed storage, and a pause does not touch it. A paused cluster holding 5 TB keeps paying for 5 TB. Older DC2 clusters bundle storage into the node price, so their pause savings look proportionally larger, but DC2 fleets are usually migration candidates anyway.
The morning resume is minutes, then cold
Resume typically takes several minutes before the cluster accepts connections, and the first queries after it compile fresh and hit cold caches, so dashboards land slower at 8:05 than at their mid-day norm. Schedule the resume 15-30 minutes before the first analyst arrives and let automated warm-up queries take the hit instead of a person.
Result caching softens the landing for repeated dashboard queries, but only after each query has taken its first cold run of the day.
Two configurations that defeat the schedule
Reserved nodes bill for their full term whether the cluster is paused or running, so pausing a fully reserved cluster saves zero. And Redshift Serverless has no pause at all because it already bills per RPU-second used; putting a calendar on it adds nothing. The pause lever belongs to on-demand provisioned clusters.