aws_redshiftserverless_workgroup cost estimation

An aws_redshiftserverless_workgroup runs Redshift queries without a provisioned cluster. Compute is measured in Redshift Processing Units (RPUs) and billed per RPU-hour at ~$0.36 — but only while queries are actually executing. An idle workgroup bills no compute, which is the core difference from provisioned Redshift's always-on nodes.

100 RPU-hours of query compute is ~$36. The cost is the product of your base RPU capacity, the queries' RPU usage, and how much you query. Because it scales to zero between queries, Serverless is cheaper than a provisioned cluster for sporadic, bursty, or unpredictable analytics — and more expensive for constant, heavy querying where a reserved provisioned cluster amortizes better.

The levers are base capacity (the RPU floor a query starts at — higher means faster but more RPU-hours), query efficiency (partitioning and good SQL reduce RPU consumption), and a usage limit to cap runaway spend.

c3x prices the workgroup from monthly_rpu_hours since compute is entirely query-driven; set it to model expected usage.

Terraform example

A minimal but realistic configuration that C3X can estimate.

resource "aws_redshiftserverless_workgroup" "analytics" {
  namespace_name = aws_redshiftserverless_namespace.main.namespace_name
  workgroup_name = "analytics"
  base_capacity  = 32
}

Pricing dimensions

What you actually pay for when you provision aws_redshiftserverless_workgroup.

Dimension	Unit	What's being charged
Redshift Processing Units	per RPU-hour	Billed per RPU-hour only while queries run; idle workgroups bill no compute. Driven by base capacity and query volume. $0.36/RPU-hour → 100 RPU-hours ≈ $36/month
Managed storage	per GB-month	Data stored in the namespace bills separately per GB-month (Redshift managed storage rate).

Sample C3X output

100 RPU-hours of query compute in a month:

aws_redshiftserverless_workgroup.analytics
└─ Redshift Processing Units   100 RPU-hours   $36.00
                               Monthly         $36.00

Optimization tips

Common ways to reduce aws_redshiftserverless_workgroup cost without changing the workload.

Let it scale to zero

Full compute cost during idle periods

The advantage over provisioned Redshift is paying nothing for compute when idle. Don't add synthetic keep-alive queries; let the workgroup go quiet between real workloads.

Right-size base capacity

Proportional to base capacity

Base capacity (RPUs) sets the floor a query starts at — higher is faster but burns RPU-hours faster. Tune it to the query mix; small/ad-hoc workloads rarely need a high floor.

Set a usage limit

Caps worst-case spend

Redshift Serverless usage limits cap RPU-hours per period, preventing a runaway query or dashboard from generating a surprise bill.

Compare against a provisioned cluster at steady load

Workload-dependent

For constant, heavy querying, a reserved provisioned Redshift cluster can be cheaper than per-RPU-hour serverless. Re-evaluate if the workgroup is rarely idle.

FAQ

How is Redshift Serverless billed?

Per RPU-hour (~$0.36) only while queries execute — an idle workgroup bills no compute. Managed storage bills separately per GB-month. Cost is base capacity × query RPU usage × how much you query.

Is Serverless cheaper than provisioned Redshift?

For sporadic, bursty, or unpredictable analytics, yes — it scales to zero between queries. For constant, heavy querying, a reserved provisioned cluster usually amortizes better. The crossover is how often the workgroup is actually idle.

How does c3x estimate the cost?

From monthly_rpu_hours, since compute is entirely query-driven. Set it to your expected RPU-hours to model the bill; storage is priced separately.

Related resources

AWS

aws_redshift_cluster

Provisioned Redshift — cheaper for constant heavy querying

`aws_redshiftserverless_workgroup` cost estimation