Skip to main content

Pricing for Aiven for Apache Kafka®

You pay for the Kafka service plan, retained data, and traffic through your topics. Aiven bills these as compute, storage, and network usage.

Pricing components​

Aiven bills the following components separately:

  • Compute: The cost of the selected service plan.
  • Storage: The cost of retained data, based on the amount stored and how long it is retained.
  • Network usage: The cost of data produced to and consumed from Kafka topics.

Network usage​

Network usage depends on Kafka topic traffic. Aiven measures data that producers write to Kafka topics as ingress and data that consumers read from Kafka topics as egress.

Classic topics and diskless topics can exist in the same service. Aiven measures network usage separately for each topic type. Ingress and egress can have different rates depending on whether the traffic is for classic topics or diskless topics.

Measured ingress and egress are higher than the size of the payload data your application sends and receives. Kafka protocol overhead, including record headers, message framing, and retries, can affect measured traffic, as can client batching behavior. Different client defaults can significantly affect how data is produced and consumed from the cluster.

Why egress and ingress differ​

Aiven measures the ingress and egress independently. Egress is typically higher, because data is usually read by more than one client, or consumed more than once. Reading the same data with multiple consumer groups, repeated reads, retries, or client reconnects all affect cluster egress.

Your client library and client configuration can also affect traffic. For example, a client that fetches the same data more than once, uses small fetches, or reconnects frequently can generate more consumed data than expected.

Cost estimates​

When you create a service, Aiven provides a monthly cost estimate based on your selected configuration. Compute sizing is approximated based on network ingress. This approximation relies on several assumptions, so some workloads can require more compute capacity, such as:

  • High egress fan-out patterns
  • A high number of partitions per broker
  • Uneven distribution of throughput among partitions
  • A high number of client connections
  • Enabled integrations, such as Datadog or the consumer lag predictor
  • Other factors that increase demand on compute

For more information about factors that increase broker resource usage, see Optimizing resource usage.

Review the estimated monthly cost in Service summary. For the Console steps, see Create an Aiven for Apache Kafka® Professional tier service.

View usage​

The Aiven Console shows usage information for the current billing period.

To review usage, open the service and go to Overview > Service usage. You can review:

  • Ingress and egress usage
  • Usage split by classic topics and diskless topics
  • Storage usage
  • Predicted usage for the billing period
note

The Aiven Console shows usage values in the unit that best fits the size of the number, for example bytes, KB, MB, GB, or TB. The unit can change as usage grows during the billing period.

Aiven bills storage by the GB-month. One GB-month means 1 GB of data stored for one month. The unit combines size and time. Storing 100 GB for a full month is 100 GB-months, and storing 100 GB for half a month is 50 GB-months.

Storage usage is prorated by the hour, so the Aiven Console can show usage in smaller units. For example, $0.12 per GB-month is equivalent to $0.000164 per GB-hour, and usage can appear in units such as KB-hours or MB-hours.

The Aiven Console shows network usage separately for classic topics and diskless topics. Predicted usage is based on your usage so far in the billing period. Usage information shown during the billing period can change as Aiven processes new usage data.

Cost drivers​

The following factors affect your estimated or actual cost:

  • Cloud and region: Prices vary by region.
  • Service plan: Determines the Kafka cluster deployed and the compute rate.
  • Topic type: Classic topics and diskless topics have different ingress and egress rates, so the share of traffic that uses each topic type affects network usage costs. For network usage rates, see the Aiven pricing page.
  • Data produced and read: Data produced and consumed is billed.
  • Retention period: Longer retention keeps data in storage for longer.

Manage costs​

To manage costs, review the factors that affect your compute, storage, and network usage:

  • Review the service plan that matches your workload. Consult an expert for any tuning.
  • Use diskless topics for high-throughput workloads that do not require low latency.
  • Adjust topic retention to control storage usage.
  • Review consumer applications and client configuration if egress is higher than expected.
  • Monitor usage during the billing period to identify unexpected changes.

Pricing by service type​

If Service type appears when you create a service, pricing depends on the selected type:

  • Standard: Aiven bills compute, storage, and network usage separately.
  • Classic: Aiven includes network usage in the compute price.

Related pages