Cluster Linking Billing
A cluster link connects a source cluster to a destination cluster, at least one of which is a Confluent Cloud cluster. Confluent can charge both the source and destination clusters for a given link, but charges only Confluent Cloud clusters, not external ones.
For supported cluster types, see Supported cluster types.
Cluster Linking is billed based on:
Cost per cluster link per hour
Ingress per GB written
Egress per GB read
Each of these factors is adjusted by a regional multiplier. The ingress rate also depends on the cluster type. Ingress and egress include metadata sync traffic.
For current prices, see the Confluent Cloud pricing page.
Cluster link charges
A cluster is charged for every cluster link where it’s a source or destination. The hourly rate depends on how the two clusters are networked, from lowest to highest:
Public or internet networking
Same-cloud private networking
Cross-cloud private networking
Cross-cloud private networking applies when the link connects clusters in different cloud providers that both use Confluent Cloud private networking, and premium regions cost more. To check whether a region is premium, see the “Premium region (cross-cloud private networking)” row in the region availability tables at Confluent Cloud Regions and Availability by Cloud Provider.
A bidirectional link is charged as two links. Both are billed at the higher of the two clusters’ rates. If a cluster link connects a Confluent Cloud cluster with a non-Confluent Cloud cluster, only the Confluent Cloud cluster is charged. Although Confluent doesn’t charge the external cluster, its own cloud provider might.
You can create multiple cluster links with the same source cluster, destination cluster, and security configuration; Confluent doesn’t de-duplicate them, so each is billed separately. Confluent Cloud charges for a cluster link even with no mirror topic or data throughput.
Ingress and egress
Confluent bills Cluster Linking data as ingress and egress on each Confluent Cloud
cluster that sends or receives it. The ingress line item is
CLUSTER_LINKING_WRITE and the egress line item is CLUSTER_LINKING_READ.
Billing includes metadata syncs and applies
regardless of networking type. These rates are separate from standard Apache Kafka®
ingress and egress, so you aren’t also charged standard Kafka ingress or egress
for this traffic. Between two Confluent Cloud clusters, a cloud provider doesn’t
charge you for this ingress. For egress, Confluent pays the cloud
provider’s egress cost for all data a Confluent Cloud cluster sends over a cluster
link.
Volume discounts are available for high-throughput use; contact your Confluent Cloud account team to find out more.
Metadata syncs
Cluster Linking metadata syncs cover topic configurations, consumer group offsets, and access control lists, billed at the same rate as topic message data. In addition to the message data mirrored from source topics, Cluster Linking syncs this metadata, so total cluster link throughput might exceed the produce throughput on source topics.
To measure the amount of metadata throughput, subtract
io.confluent.kafka.server/cluster_link_mirror_topic_bytes, which accounts
only for topic data, from
io.confluent.kafka.server/cluster_link_destination_response_bytes, which
accounts for all cluster link data.
Cost management
You can reduce what a cluster link costs by:
Synchronizing less frequently. A longer consumer.offset.sync.ms interval reduces metadata throughput and cost. Set it based on how quickly you need offsets to catch up.
Using unidirectional links when you don’t need bidirectional replication. A bidirectional link is billed as two links, both at the higher of the two clusters’ hourly rates.
Securing volume discounts if you run Cluster Linking at high throughput. See Annual commitments.
For worked pricing examples covering active-passive, active-active, and hybrid cloud/migration configurations, see Cluster Linking.