ApsaraMQ for Kafka:Pay-as-you-go billing

Pricing example

A VPC-connected Professional Edition (High Write) instance in Region group 1 with the following configuration:

Hourly cost:

Traffic specification

The traffic specification determines your cluster's compute and processing capacity, measured by ENI throughput:

ENI traffic = Business traffic + Replication traffic

• Business traffic: The actual messaging throughput generated by producers and consumers.

• Replication traffic: The throughput consumed by data replication across the default 3 replicas.

Read/write ratios by edition:

To choose the right traffic specification, compare the peak values of read and write traffic, then select a specification based on the higher value. Purchase approximately 130% of the peak value to create a buffer for traffic spikes.

For supported traffic specifications per edition, see Instance editions.

Disk capacity

Disk capacity stores your messages and their replicas.

• Minimum capacity: Varies by traffic specification to balance performance and cost.

• Disk types: Ultra disks and SSDs. SSDs deliver better performance.

• Irreversible choice: The disk type cannot be changed after purchase.

• 3 replicas by default: All data is stored in 3 replicas.

Usable storage by edition:

The disk cost of a Professional Edition instance with 3 replicas is approximately 60% lower than a self-managed Apache Kafka cluster.

Manage disk usage: Adjust the Message Retention Period parameter to control how long messages are kept. The default retention is 72 hours, configurable from 24 to 480 hours. When disk usage reaches 85%, the system automatically deletes the oldest messages to maintain availability. For details, see Modify message configurations. For more information about message deletion, see When are historical messages deleted from ApsaraMQ for Kafka?

Partitions

Partitions are the basic storage unit within a topic. Each traffic specification includes a set of free partitions.

How partition counts work:

Total partitions = Free partitions (from traffic specification) + Purchased partitions
Available topics  = Total partitions
Available groups  = Total partitions x 2

Example: With the alikafka.hw.2xlarge traffic specification (1,000 free partitions) and 50 purchased partitions:

• Total partitions: 1,050

• Available topics: 1,050

• Available groups: 2,100

For the number of partitions by instance type, see Instance editions.

Topic-based purchasing is no longer available. Existing instances purchased by topic specification have been automatically converted to partition-based specifications. The number of partitions after conversion is based on the number of topics that you purchased.

Internet traffic

Internet traffic is billed only for Internet- and VPC-connected instances. Pricing is based on elastic IP address (EIP) bandwidth using BGP (Multi-ISP) lines. Calculation results are rounded to three decimal places.

Internet traffic pricing

Unit: USD per hour.

Traffic specification pricing

Prices vary by instance edition and region group. For region group definitions, see Region groups.

For traffic specifications supported by instances purchased based on the topic specification, see Free topics and partitions for topic-based instances.

Disk capacity pricing

Prices vary by disk type and region group. For region group definitions, see Region groups.

Unit: USD per hour.

The disk cost of a Professional Edition instance with 3 replicas is approximately 60% lower than a self-managed open source Apache Kafka cluster.

How do I estimate the storage cost change after adjusting the message retention period?

You can use the following steps to estimate how adjusting the Message Retention Period parameter affects your disk storage costs.

Step 1: Estimate your storage requirement

Estimated storage (GB) ≈ Daily average write volume (GB/day) × Target retention period (days)

Step 2: Calculate the estimated disk cost

Disk cost (per day) = Estimated storage (GB) × Disk unit price (per GB per hour) × 24

For the disk unit price, refer to the disk capacity pricing tables above. For example, Region group 1, ultra disk: USD 0.01 per 100 GB per hour (USD 0.0001 per GB per hour).

Step 3: Calculate the savings

Estimated savings = Cost at original retention period − Cost at new retention period

Example: Your instance writes 50 GB of data per day. You plan to reduce the message retention period from 3 days to 2 days.

Partition pricing

If the free partitions included in your traffic specification are not enough, purchase additional partitions. Prices vary by region group.

Unit: USD per partition-hour.

For partition limits by instance type, see Instance editions.