All Products
Search
Document Center

ApsaraMQ for Kafka:Billing FAQ

Last Updated:Mar 10, 2026

Why am I charged for a pay-as-you-go instance I have not used?

Pay-as-you-go instances are billed based on how long they exist, not on message volume or consumer activity. When you create an instance, dedicated brokers and storage are immediately allocated and billing starts. Charges continue until you release the instance.

Billing formula

Total fees = Instance price x Service time

For example, if your instance runs for 3 days before you release it, you pay for 72 hours of service time, regardless of whether any messages were produced or consumed.

How to stop charges

Release the instance. The following actions do not stop billing:

  • Stopping producers and consumers

  • Deleting topics

These actions leave the underlying broker and storage resources allocated, so charges continue to accrue.

Can I pause billing without releasing the instance?

No. There is no way to pause or suspend billing. As long as the instance exists, you are charged.

How is a topic-based instance converted to a partition-based instance?

You can no longer purchase ApsaraMQ for Kafka instances by topic specification. All instances now use partition specification.

Regardless of whether your instance was originally purchased by topic or partition specification, upgrades and downgrades now use partition specification only. When you upgrade or downgrade a topic-based instance, it is automatically converted using the following rules.

Conversion formula

Total partitions after conversion = Partitions included in the specification + Additional topics purchased x Conversion coefficient

The result is rounded up to the nearest integer.

Conversion coefficients

Billing methodConversion coefficient
Subscription23.913
Pay-as-you-go45.045

Worked example

A subscription Standard Edition instance has the following configuration:

  • Bandwidth: 20 MB/s

  • Topics in the specification: 50

  • Additional topics purchased: 10

After conversion:

1,000 + 10 x 40.602 = 1,407 partitions

For partition quotas included in each specification, see Billing rules.

Shortfall protection

If your actual partition usage exceeds the converted total, Alibaba Cloud provides the extra partitions at no charge. For example, if the conversion yields 1,000 partitions but you are using 1,100, Alibaba Cloud supplements the 100-partition difference free of charge.

Post-conversion quotas

After conversion, topic and group quotas are updated based on the partition count:

ResourceNew quota
TopicsEqual to the total number of partitions
GroupsEqual to the total number of partitions x 2

Why am I charged by peak traffic?

Generally, billing is either hardware-based or service-based. In the hardware-based billing mode, it is difficult to bill an instance based on actual service usage. In the service-based billing mode, the instance is billed based on actual service usage.

What does peak traffic measure?

Peak traffic defines the overall data processing capability of your instance, not just network bandwidth. Selecting a peak traffic tier provisions the throughput capacity of the entire cluster.

Daily processing capability

Calculate the minimum daily processing capability from peak traffic:

Data processing capability = Peak traffic x Duration

Example: An instance with 30 MB/s peak traffic can process at least 2.47 TB per day:

30 MB/s x 60 s x 60 min x 24 h / 1024 / 1024 = 2.47 TB/day
Peak trafficMinimum daily processing capability
20 MB/s~1.65 TB/day
30 MB/s~2.47 TB/day
60 MB/s~4.94 TB/day
120 MB/s~9.89 TB/day
Actual processing capability may exceed these minimums.

Why am I charged for additional partitions?

Open-source Apache Kafka clusters suffer significant performance degradation beyond about 1,000 partitions. ApsaraMQ for Kafka eliminates this limitation -- a topic can have an unlimited number of partitions without performance loss.

However, additional partitions beyond your resource plan consume broker resources, so you are billed for these extra partitions accordingly.

How does the connector feature affect my bill?

The connector feature is built on EventBridge and Function Compute. Both services appear as separate line items on your bill:

ServiceBilling status
EventBridgeIn public preview, free of charge
Function ComputeBilled separately. See Billing overview.
This applies to the new version of the connector feature. For the previous version, see Overview.