Billing Instructions for BYOC
Reference to the billing overview, this article provides a detailed explanation of each billing item involved in the AutoMQ Cloud product under the BYOC environment.
The terms such as cloud providers and Public Cloud providers mentioned in this article refer to mainstream cloud service providers, such as AWS, Google Cloud, Azure, Alibaba Cloud, Tencent Cloud, Huawei Cloud, etc.
AutoMQ Subscription Fees (Payable to AutoMQ)
When using AutoMQ Cloud in a BYOC environment, subscription fees must be paid based on the actual cluster size. The metrics to measure subscription fees for AutoMQ include:
- Message Processing Specification AKU (AutoMQ for Kafka Unit): Required. The AKU measures the scale of computational processing resources allocated during the sending and receiving of messages in a Kafka cluster. The AKU specification is positively correlated with the cluster's scale pressure.
Message Processing Specification AKU Fees
Indicator Constraints
The Message Processing Specification AKU represents the computational processing resources allocated during message transmission within the cluster. The AKU considers resource consumption such as computational processing, storage IOPS, and network throughput.
Allocating appropriate AKU specifications for each AutoMQ instance ensures the required message transmission throughput capability. Based on benchmark performance test results, each additional AKU specification provides the following performance capabilities:
Capabilities Provided by Each 1 AKU | Description | |
---|---|---|
Message Write Throughput | 20 MB/s | Core business metrics and pressure for message read/write calls. Exceeding the predetermined specifications may cause slower service response, increased RT, or throttling failures. |
Message Read Throughput | 20 MB/s | |
Client Request Frequency | 800 requests/second | The request frequency and pressure to the server accessed via Kafka Producer and Consumer SDK. If it exceeds the predetermined specifications, it may lead to throttling or increased response time (RT). Request types include:
|
Maximum number of partitions | 1125 | The number of partitions constrains the scale of metadata that an instance can simultaneously hold. If the predefined specifications are exceeded, the following issues may arise:
AutoMQ's S3-based shared storage architecture supports 10 times the partition performance compared to Apache Kafka for the same cluster size without the risk of performance degradation due to too many partitions. If the default number of partitions does not meet your needs, please contact us through Obtaining Services▸. |
In a production environment, it is recommended that applications strictly follow the above capacity specifications to evaluate resource consumption and perform timely scaling to avoid excessive pressure on the cluster, which could affect service stability.
Calculation Rules (example of Hourly Billing)
Each AutoMQ instance (cluster) can specify the required AKU specifications during creation or specification changes. During billing, the number of AKUs consumed in real-time by each instance will be accounted for.
Scope of statistics: The number of AKUs consumed by each AutoMQ Kafka instance.
Statistical method: Calculate the peak value within an hourly cycle.
Aggregation method: Aggregate statistics based on instance granularity.
Measurement unit: AKU·Hour, where actual consumption is measured on an hourly basis, with any period less than an hour counted as a full hour.
Billing Example (hourly Billing)
Based on the above calculation rules, here are some typical billing scenarios:
Example 1:
A user creates an instance at 10:00 with a capacity of 12 AKUs. At 11:00, the user scales up the instance to 24 AKUs.
From 10:00 to 12:00, the user's actual AKU consumption is:
12 * 1 hour + 24 * 1 hour = 36 AKU·hours.
The above example uses real-time billing in the cloud market, with subscription fees calculated on an hourly basis. If opting for monthly billing or purchasing a subscription license in advance, you will also need to measure the number of AKUs consumed by each instance.
Cloud Resource Costs (cloud Resources on Which Deployed Instances Depend, Payable to Cloud Providers)
Deploying an AutoMQ instance incurs resource consumption in addition to subscription fees. For example, using Alibaba Cloud as a Cloud provider:
Cloud Product | Specifications and Usage | Dependency Description |
---|---|---|
ECS (Elastic Compute Service) |
|
|
Cloud Disk (Elastic Block Service) |
|
|
Object Storage |
|
|