Skip to Main Content

Billing Instructions for BYOC

The billing overview section offers detailed information about the billing components associated with the AutoMQ Cloud product in a BYOC (Bring Your Own Cloud) environment.

The terms cloud vendors and public cloud vendors mentioned in this document refer to major cloud service providers such as AWS, Google Cloud, Azure, Alibaba Cloud, Tencent Cloud, Huawei Cloud, etc.

AutoMQ Subscription Fees (Payable to AutoMQ)​

To use AutoMQ Cloud in a BYOC environment, users are required to pay subscription fees based on the actual cluster size. AutoMQ evaluates subscription fees with the following scale indicators:

  • Message Processing Specification AKU (AutoMQ Kafka Unit): Mandatory, the AKU message processing specification is used to measure the scale of computing resources allocated during the sending and receiving of messages within a Kafka cluster. The AKU specification is directly proportional to the cluster's scale pressure.

Message Processing Specification AKU Costs​

Metric Constraints​

The Message Processing Specification AKU refers to the computational processing resources allocated within a cluster during message transmission. AKU considers resource consumption factors such as computational power, storage IOPS, and network throughput.

Allocating the appropriate AKU specifications for each AutoMQ instance ensures that the required message transmission throughput capacity is achieved. According to benchmark performance test results, adding 1 AKU specification delivers the following performance capabilities:

Capabilities Provided by 1 AKU
Description

Read/Write Throughput


Write 30 MiB/s or Read 60 MiB/s


Core performance metrics and load for message read/write calls. If the predefined specifications are exceeded, it may result in slower service response times, increased RT, or rate-limiting failures.



Example: For a specific AutoMQ instance, the write throughput is 60 MiB/s and the read throughput is 240 MiB/s. This instance's write throughput consumes 60/30 = 2 AKU, read throughput consumes 240/60 = 6 AKU, totaling 8 AKU.



Client Request Rate


800 requests per second


The application interacts with the server using the Kafka Producer and Consumer SDK, managing request rate and load. If the rate surpasses the predefined specifications, it may lead to throttling or longer response times (RT).


Request types include:


  • Produce
  • FetchConsumer
  • CommitOffset

Upper limit on partition count


1125 pieces


The number of partitions limits the amount of metadata an instance can handle simultaneously. Exceeding the specified limit may result in the following situations:


  • Inability to create new topics or expand partitions.

AutoMQ, using an S3-based shared storage architecture, compared to Apache Kafka at the same cluster scale, supports 10 times Partition performance, without concerns about performance degradation due to an excessive number of partitions.


If the default number of partitions is insufficient for your requirements, please Obtaining Services▸ contact us.



In a production environment, it is recommended that applications strictly adhere to the aforementioned specifications for processing capacity to evaluate resource consumption. This will allow for prompt scaling up or down as needed. This approach helps prevent excessive usage from overloading the cluster, which could impact service stability.

Calculation Rules​

Each AutoMQ instance (cluster) allows you to specify the desired AKU specifications when creating or modifying its specifications. The billing system records the real-time consumption of AKU numbers for each instance.

  • Scope of Statistics: The count of AKUs for each AutoMQ Kafka instance.

  • Statistics Method: An hourly cycle is used to record the peak value within each cycle.

  • Aggregation Method: Statistics are aggregated at the instance level.

Deploying an AutoMQ instance involves resource consumption costs in addition to the subscription fee. Using Alibaba Cloud as an example:

Cloud Product
Specifications and Usage
Dependency Description
VM (Virtual Machine)
  • Specifications: Varies depending on the region.
  • Usage: Scales with the size of the user cluster.
  • AutoMQ Kafka's VM consumption changes with the size of the user cluster.
  • At least 1 VM is consumed to deploy the management interface even if no instance is created.
Elastic Block Service (Cloud Disk)
  • Specifications: PL1 type
  • Usage: Each Kafka data node uses 40GB.
  • AutoMQ Kafka uses cloud disks to store temporary message data.
Object Storage
  • Specifications: Standard disaster recovery storage
  • Usage: Two buckets per environment; storage capacity scales with user environment resource consumption.
  • Core storage dependency of the message cluster.
  • Charges to cloud provider are based on actual space usage and call volume.