AWS Redshift Support
Redshift Serverless vs provisioned clusters.
Serverless bills for compute only while queries run, measured in Redshift Processing Units (RPUs). Provisioned RA3 clusters bill for nodes as long as they run, with reservations for steady use. Which is cheaper depends on how busy the warehouse really is. We measure first, then recommend.
When You Need This
The situations where this work pays for itself.
Signs it is time
- Your Redshift cluster sits idle for much of the day
- Workloads are spiky or unpredictable
- Bills are high and you suspect over-provisioning
- Several teams need isolated compute against shared data
- You are choosing for a new warehouse
What you get
- Utilization analysis and cost comparison
- Recommendation with the reasoning
- Migration runbook and rollback plan
- Serverless capacity limits and usage alarms
How we do it
A clear, safe process you can follow along with.
1. Measure how busy the warehouse is
We analyze CPU, query counts, concurrency and queue times by hour using CloudWatch and system views, to see how many hours the cluster does real work.
2. Model both options
We compare provisioned node hours, with and without reserved nodes, against Serverless RPU-hours for the same workload, using current regional prices and your measured pattern.
3. Consider workload shape
Spiky and intermittent use usually favors Serverless. Steady heavy load running most of the day often favors provisioned with reservations. We also weigh feature differences and limits, which we verify for your use case.
4. Plan the migration
Moving uses a snapshot restore to Serverless or the reverse. We test performance on the target, including your slowest queries, before switching.
5. Set capacity and guardrails in Serverless
We set base capacity, a maximum RPU level and usage limits that alert or stop spending, so a runaway query cannot generate a surprise bill.
6. Cut over and review
We switch connection endpoints, monitor for several weeks and adjust capacity from observed query times and cost.
Mistakes we help you avoid
Common ways this goes wrong, and what we do instead.
Choosing Serverless for an always-busy warehouse
Continuous heavy use can cost more than reserved nodes.
Setting no usage limit
One bad query can run up cost. Use limits and alarms.
Skipping a performance test
Query times may differ on the new compute and should be checked first.
Frequently Asked Questions
Is Redshift Serverless cheaper than provisioned?
For intermittent or unpredictable workloads often yes, because idle time is not billed. For steady, heavy use provisioned with reservations can be cheaper. Model your own usage.
What is an RPU?
A Redshift Processing Unit is the unit of Serverless compute capacity. You set a base capacity and optionally a maximum, and you pay for RPU time used.
Can I move from provisioned to Serverless?
Yes, you can restore a provisioned cluster snapshot into a Serverless namespace. Test queries and permissions before you cut over.
Want Help With Redshift Serverless vs provisioned clusters?
Schedule an appointment. Tell us about your environment and we will outline the safest, highest-value approach.
Schedule Appointment › No commitment required