AWSAmazon RedshiftAnalytics

`aws_redshift_parameter_group` cost estimation

A set of engine parameters for a Redshift cluster. Parameter groups are free. The clusters they configure bill per node-hour, from $0.25/hour for dc2.large to $13/hour for ra3.16xlarge, plus Spectrum and concurrency scaling.

The aws_redshift_parameter_group resource holds cluster-level settings (workload management queues, statement_timeout, enable_user_activity_logging, require_ssl, and more) that you attach to a Redshift cluster. Parameter groups are free. There is no charge for creating one or attaching it.

The cost is the cluster. Redshift provisioned clusters bill per node-hour by node type. A dc2.large is about $0.25/hour (roughly $180/month per node), an ra3.xlplus about $1.086/hour, an ra3.4xlarge about $3.26/hour, and an ra3.16xlarge about $13.04/hour. Clusters have multiple nodes, so a 4-node ra3.4xlarge is around $13/hour, roughly $9,500/month on demand before storage and add-ons. RA3 nodes bill managed storage separately at about $0.024/GB-month, which decouples storage from compute. Redshift Serverless is the alternative, billing about $0.36 to $0.45 per RPU-hour with an 8 RPU minimum, better for intermittent workloads than an always-on cluster.

The parameter group is cost-adjacent through workload management and logging. The WLM configuration in the parameter group governs query concurrency and memory, which directly influences whether concurrency scaling kicks in. Concurrency scaling adds transient capacity to handle bursts and is billed per second beyond a free daily credit (one hour of free concurrency scaling credit accrues per day per cluster); a poorly tuned WLM that constantly spills to concurrency scaling generates real extra cost. The enable_user_activity_logging parameter turns on audit logs that land in S3 or CloudWatch Logs, adding storage and ingestion charges. Parameters that affect query efficiency (statement_timeout, WLM memory allocation) indirectly change how much Spectrum data you scan, and Redshift Spectrum bills $5 per TB scanned against S3.

A gotcha: parameter changes often require a cluster reboot to take effect, and the default parameter group cannot be modified, so tuning WLM or enabling SSL requires a custom group (this free resource). Over-provisioning is the dominant Redshift cost mistake: clusters sized for peak run 24/7 at peak cost, so pausing idle clusters or moving bursty analytics to Serverless is usually the biggest lever.

c3x flags aws_redshift_parameter_group as free and attributes node-hour, storage, concurrency scaling, and Spectrum cost to the cluster.

Terraform example

A minimal but realistic configuration that C3X can estimate.

resource "aws_redshift_parameter_group" "main" {
  name   = "redshift-wlm-tuned"
  family = "redshift-1.0"

  parameter {
    name  = "require_ssl"
    value = "true"
  }

  parameter {
    name  = "statement_timeout"
    value = "600000"
  }

  parameter {
    name = "wlm_json_configuration"
    value = jsonencode([
      { query_concurrency = 5, memory_percent_to_use = 60 },
      { query_concurrency = 3, memory_percent_to_use = 40 },
    ])
  }
}

resource "aws_redshift_cluster" "analytics" {
  cluster_identifier        = "analytics"
  node_type                 = "ra3.xlplus"
  number_of_nodes           = 2
  cluster_type              = "multi-node"
  database_name             = "analytics"
  master_username           = "admin"
  manage_master_password    = true
  parameter_group_name      = aws_redshift_parameter_group.main.name
  publicly_accessible       = false
}

Pricing dimensions

What you actually pay for when you provision aws_redshift_parameter_group.

Dimension	Unit	What's being charged
Parameter group	free	Creating and attaching custom parameter groups carries no AWS charge. $0 (free)
Cluster node	per hour	The cluster bills per node-hour by node type, multiplied by node count. $0.25/hour (dc2.large) to $13.04/hour (ra3.16xlarge)
Concurrency scaling	per second	Transient burst capacity billed per second beyond a free daily credit. WLM tuning in the parameter group affects how often it triggers. Billed at the per-second on-demand node rate after the daily free hour
Redshift Spectrum	per TB scanned	Queries over external S3 data bill by bytes scanned; query efficiency settings influence scan volume. $5.00 per TB scanned

Optimization tips

Common ways to reduce aws_redshift_parameter_group cost without changing the workload.

Pause or move idle clusters to Serverless

Up to 65 percent for a business-hours-only workload

A provisioned cluster bills 24/7 even when nobody queries it. Pausing clusters outside business hours or moving intermittent analytics to Redshift Serverless (8 RPU minimum, billed per RPU-hour) removes the always-on cost.

Tune WLM to avoid unnecessary concurrency scaling

Concurrency scaling bills per second past the daily free hour. A WLM configuration in the parameter group that allocates queues and memory sensibly keeps normal load on the base cluster and reserves scaling for genuine bursts.

Prefer RA3 to decouple storage from compute

RA3 nodes bill managed storage separately at about $0.024/GB-month, so you size compute for query load rather than for data volume. For large datasets on older DC2 nodes, this often lets you run fewer, smaller compute nodes.

Limit Spectrum scan volume

Parquet plus partitioning can cut scan cost 10x

Spectrum bills $5 per TB scanned. Partitioning external data, using columnar formats like Parquet, and pushing predicates down cut the bytes scanned per query dramatically.

FAQ

Do Redshift parameter groups cost anything?

No. Parameter groups are free. You pay for the cluster nodes (per node-hour), managed storage on RA3, concurrency scaling beyond the daily free credit, and any Spectrum data scanned. The parameter group only tunes how efficiently the cluster runs.

How does a parameter setting affect concurrency scaling cost?

The WLM configuration lives in the parameter group and controls query concurrency and memory. If WLM is set so queries constantly queue and spill, Redshift triggers concurrency scaling more often, and that burst capacity bills per second past the free daily hour. Better WLM tuning keeps load on the base cluster.

Is Redshift Serverless cheaper than a provisioned cluster?

For intermittent or bursty workloads, usually yes, because Serverless bills per RPU-hour only while running instead of a cluster billing 24/7. For a steady, heavily used cluster with good utilization, provisioned (especially with reserved nodes) is typically cheaper.

Why create a custom parameter group?

The default parameter group cannot be modified. To tune WLM, enable SSL, set statement timeouts, or turn on user activity logging, you create a custom group (this free resource) and attach it. Those settings are what let the cluster run efficiently and avoid needless burst charges.