# Amazon Redshift Cluster

> A provisioned Redshift cluster bills per node-hour for every node, around the clock, and RA3 managed storage adds a per-GB-month meter. ZopNight pauses non-production clusters on schedule via redshift:PauseCluster, which stops the compute meter while storage keeps billing, and resumes them via redshift:ResumeCluster.

Source: https://zop.dev/integrations/aws/redshift
Updated: 2026-08-19

---

Amazon Redshift is a managed data warehouse billed per node-hour for provisioned clusters. Warehouses used for business-hours analytics but left running 24/7 pay for empty nights and weekends.

## Node-hours, times every node

A provisioned cluster bills each node's hourly rate around the clock, so an 8-node cluster is eight meters running in parallel. RA3 node types bill managed storage per GB-month separately from compute; older DC2 nodes bundle storage into the node price. Concurrency scaling and Spectrum queries meter on their own when used, but for most clusters the node-hour product is the bill.

## Pause is a real stop verb here

Unlike most databases, Redshift has a native pause: redshift:PauseCluster halts the per-node compute meter while keeping the cluster's data in place, and redshift:ResumeCluster brings it back in minutes. Storage keeps billing while paused. ZopNight drives exactly this pair on schedule, which fits the dominant Redshift usage shape: analysts query from 9 to 6, and the warehouse serves nothing overnight or on weekends.

## Where warehouse money leaks

The classic pattern is the BI cluster running 168 hours a week for a 45-hour query week; a pause schedule recovers most of the difference. Next come clusters sized for a quarterly peak and never scaled back down, and long-lived steady clusters paying On-Demand node rates that a reserved-node purchase would cut substantially.

## How ZopNight covers Redshift

Clusters are found on the 6-hour cycle. Hourly CloudWatch metrics from the Redshift namespace are kept with a 90-day lookback, and per-cluster cost comes from Cost Explorer or CUR 2.0. Recommendations include off-hours scheduling, node rightsizing, and reserved-node opportunities.

## Pause lives in the console too

Redshift console, then Clusters. The list shows node type and node count; the Actions menu on a cluster includes Pause and Resume, the same operations a schedule automates, so anything done by hand there is visible to and reconciled with the scheduler.