This topic describes the resource metering method, billable items, and billing formulas of fully managed Flink. This topic also describes the costs of other cloud services that are required to deploy and use fully managed Flink. You can refer to this topic to estimate the resources and costs required to deploy fully managed Flink for your business.
The billing unit of fully managed Flink is compute unit (CU), which is a unit of measurement for computing resources. One CU is equal to 1 CPU core, 4 GiB of memory, and 20 GB of local storage. The local storage stores information, such as logs and checkpoints. The number of CUs consumed reflects the resources used by Realtime Compute for Apache Flink. The number of CUs consumed by a Realtime Compute for Apache Flink deployment varies based on the queries per second (QPS) of input data streams, computing complexity, and input data distribution of the deployment. You can estimate the number of CUs that you need to purchase based on your business scale and the computing capability of Realtime Compute for Apache Flink. The following table describes the processing capability of CUs in Realtime Compute for Apache Flink.
Simple streaming stress testing
The operations include filtering and cleansing.
Each CU can process 40,000 to 55,000 data records per second.
Complex streaming stress testing
The operations include aggregation and complex user-defined function (UDF) calculation.
Each CU can process 5,000 to 10,000 data records per second.
The preceding estimate applies only to the internal processing capability of Realtime Compute for Apache Flink. The external data read and write capabilities are not included. The external data read and write efficiency may affect the computing capability estimate of Realtime Compute for Apache Flink.
If you want to use Realtime Compute for Apache Flink to read data from Simple Log Service but the query quota of Simple Log Service is limited, the overall computing capability of Realtime Compute for Apache Flink is subject to the capability allowed by Simple Log Service.
If the number of connections or transactions per second (TPS) is limited for the ApsaraDB RDS database that Realtime Compute for Apache Flink references, the throughput of Realtime Compute for Apache Flink is limited by the throttling of the ApsaraDB RDS database.
Window functions are complex operations and consume more CUs than simple operations. If your applications require the use of window functions, we recommend that you purchase at least four CUs.
The following table describes the billable items of fully managed Flink.
When you create a workspace, the system deploys a console for the workspace. The console and its components consume about 2 CUs for management.
Computing resources are the resources that are consumed when you perform computing tasks, and are metered in CUs. You are charged for the CUs based on the subscription or pay-as-you-go billing method.
The following table describes how the fees for each billing method are calculated for individual workspaces.
Workspace fee = (CUs consumed for management + CUs consumed for computing) × Monthly price × Subscription period
Workspace fee = Sum of computing resource usage per minute in an hour/60 × Hourly price + Management resource usage in an hour/60 × Number of CUs consumed for management × Hourly price
A pay-as-you-go workspace is billed by minute. The billing cycle is 1 hour.
The number of CUs consumed for management is fixed to 2.
The fee calculated by using the preceding formula is the fee for one workspace of fully managed Flink. You can purchase multiple workspaces for an Alibaba Cloud account. When you settle bills for an account, you must pay fees based on the total fees of multiple workspaces.
Billing rules for supporting services
Fully managed Flink relies on several other services and can be used together with some services to extend its capabilities. Aside from the service fee for fully managed Flink, you may also be charged for other services as described in the following table.
Alibaba Cloud service
To activate fully managed Flink, you must select a virtual private cloud (VPC) in the region in which fully managed Flink resides. For more information, see Billing.
Object Storage Service (OSS) is used to store information about Flink deployments, such as checkpoints, savepoints, logs, and JAR packages. For more information, see Billing overview.
Server Load Balancer (SLB) is a necessary network connection component that is used to access the Realtime Compute for Apache Flink console by using a browser. After you activate fully managed Flink, the SLB service is automatically activated. For more information about SLB billing methods, see Pay-by-LCU (new).
After you activate a fully managed Flink workspace, Application Real-Time Monitoring Service (ARMS) is automatically activated to provide Prometheus Service Pro Edition. For more information about the billing of ARMS, see Pay-as-you-go.