Usage Metrics

Redpanda Cloud uses various metrics to measure the consumption of resources.

  • All pricing is set in US dollars (USD).

  • All billing computations are conducted in Coordinated Universal Time (UTC). Billing accrues at hourly intervals. Any usage that is less than an hour is billed for the full hour.

Pricing information is available on redpanda.com. For questions about billing, contact billing@redpanda.com.

Serverless metrics

Pricing for Serverless clusters depends on the data in, data out, data stored, partitions (virtual streams), and the time the instance is up. The cost for each Serverless metric varies based on the region you select for your cluster.

Metric Description

Uptime

Tracks the number of hours the instance is running.

Ingress

Tracks the data written into Redpanda (in GB).

All Kafka protocol requests (except message headers) are counted as ingress as soon as they are read by Redpanda’s proxy process.

Egress

Tracks the data read out of Redpanda (in GB).

All Kafka protocol responses generated by the cluster (except message headers) are counted as egress as soon as the cluster processes the request, even if the client drops the connection before they are delivered.

Partitions

Tracks the number of partitions used per hour.

Storage

Tracks the data in object storage per hour (in GB).

See also: Serverless limits

Dedicated metrics

Pricing for Dedicated clusters depends on the time the instance is up, the data in, data out, and data stored.

Metric Description

Uptime

Tracks the number of hours the instance is running.

The cost varies based on the region and tier you select for your cluster.

Ingress

Tracks the data written into Redpanda (in GB).

All Kafka protocol requests (including message headers) are counted as ingress as soon as they are read by Redpanda’s proxy process.

The cost varies based on the region you select for your cluster.

Egress

Tracks the data read out of Redpanda (in GB).

All Kafka protocol responses generated by the cluster (including message headers) are counted as egress as soon as the cluster processes the request, even if the client drops the connection before they are delivered.

The cost varies based on the number of availability zones (AZ) you select for your cluster.

Storage

Tracks the data in object storage per hour (in GB).

Replication to object storage is implemented with Tiered Storage. All topics have a fixed replication factor of 3, but Redpanda counts each byte only once.

Redpanda Connect pipeline metrics (Dedicated and Serverless)

Pricing per pipeline depends on:

  • The number of compute units you allocate.

  • The cost of a compute unit, which varies based on the cloud provider and region you select for your cluster.

Metric Description

Compute

Tracks the server resources (CPU and memory) a pipeline uses in compute units per hour.

Next steps