Realtime Compute for Apache Flink:Hybrid billing

Billing overview

Hybrid billing consists of management resources, fixed computing resources, and elastic computing resources. To use this billing method, you must first create a subscription workspace, which provides fixed computing resources. Then, you can enable hybrid billing for the workspace and set a limit for the elastic computing resources. This method uses the reserved fixed computing resources from your subscription to handle predictable and long-running tasks. The elastic computing resources provide the flexibility to manage workloads with periodic or fluctuating peak demands. For streaming deployments, this allows you to convert the resources saved by automatic tuning into cost savings. For batch deployments with distinct periodic patterns, you avoid the need to maintain high levels of computing resources long-term. This approach prevents over-provisioning for short-term peaks and mitigates the risk of resource acquisition failures when relying solely on elastic computing resources.

Usage notes

• Only Intel X86-based subscription workspaces support elastic computing resources.

• If you need to switch from hybrid billing to the subscription or pay-as-you-go method, follow these steps:

  • Switch to subscription: Disable the hybrid billing feature.

  • Switch to pay-as-you-go: First, disable the hybrid billing feature, and then switch the workspace from the subscription to the pay-as-you-go method.

• When you run a deployment, the system automatically adapts based on resource availability. If the target queue has available fixed computing resources, they are used first.

• Hybrid billing workspaces support enabling cross-zone high availability at the subscription namespace level. In case of a failure, the system prioritizes the migration and recovery of deployments that run on fixed computing resources. Deployments on elastic computing resources are recovered on a best-effort basis at no extra cost. For more information, see Cross-zone high availability.

Billing rules

Total price for a hybrid billing workspace = Fee for management resources + Fee for computing resources = Fee for management resources + Fee for fixed computing resources + Fee for elastic computing resources

• Fees for management resources and fixed computing resources: These are the fees for the management resources and computing resources in your subscription workspace. For more information, see Subscription.

• Fee for elastic computing resources: Elastic computing resources are billed based on actual usage. At the end of each billing cycle, a pay-as-you-go bill is generated, and the amount is automatically deducted from your account. The calculation method and unit price are the same as those for the pay-as-you-go method for Realtime Compute for Apache Flink. For more information, see Pay-as-you-go.

Enable and use elastic computing resources

1. Select the target subscription workspace, enable elastic computing resources, and set a limit for the elastic computing resources.

   Log on to the Realtime Compute for Apache Flink console. In the Actions column of the target subscription workspace, choose More > Enable Hybrid Billing. On the Hybrid Billing Settings page, set Elastic Resource Ceiling, click OK, and then click Pay.

   Note

   The maximum amount of resources your workspace can use is the sum of your purchased fixed computing resources and the configured elastic resource ceiling.

2. Allocate resources to the namespace and set the maximum available elastic computing resources.

   1. Click the icon to the left of the target workspace ID.

   2. In the Actions column of the target namespace, click Reconfigure Resources and set the Maximum elastic resources.

3. Create or edit a queue to allocate elastic computing resources.

   • By default, the elastic computing resources configured for a namespace are allocated to the default-queue. To make adjustments, see Manage queues.

   • You do not need to specify whether to use elastic or fixed computing resources when running a deployment. The system adapts based on your resource availability and prioritizes the use of fixed computing resources. This prevents you from being billed for elastic computing resources when the queue has idle fixed computing resources.

   • Ensure that the queue has sufficient resources to support the stable operation of deployments and avoid failures due to resource shortages.

4. Based on your business plan, deploy some deployments to a queue that has elastic computing resources. For more information, see Deployments.

Example: Using automatic tuning

You can combine hybrid billing with automatic tuning, which includes Autopilot and scheduled tuning, to convert resource savings into cost savings. For example, based on the configuration example for automatic tuning, your business has peak hours from 09:00 to 19:00 (10 hours) and off-peak hours from 19:00 to 09:00 the next day (14 hours), requiring 30 CU and 10 CU respectively.

With hybrid billing, you can purchase 10 CU (or more) of fixed computing resources for your workspace to ensure stable operation during off-peak hours. At the same time, you can enable elastic computing resources and set the maximum available elastic computing resources to 20 CU (or more). During peak hours, if the fixed computing resources are insufficient, the system uses elastic computing resources and bills you for the actual amount used, up to a maximum of 20 CU.

Compared with other billing methods, hybrid billing allows you to purchase only 10 CU of fixed computing resources in advance, which saves the cost of 20 CU of fixed computing resources. During peak periods, this method allows you to use up to 20 additional CU of elastic computing resources. The following tables compare the differences and provide cost estimations, using a workspace in Singapore for one month (30 days) as an example.

Billing method	CUs to purchase	Description	Estimated cost example
Subscription	30 CU	You must purchase resources based on peak demand to ensure that deployments run smoothly at all times.	Fee for management resources + Fee for computing resources = (2 CU × USD 45.94/CU/month × 1 month) + (30 CU × USD 45.94/CU/month × 1 month) = USD 91.88 + USD 1,378.2 = USD 1,470.08
Pay-as-you-go	Billed based on actual usage.	Resource shortages in an availability zone may prevent deployments from starting.	Estimated maximum fee: Fee for management resources + Fee for peak computing resources + Fee for off-peak computing resources = (2 CU × USD 0.09/CU/hour × 24 hours × 30 days) + (30 CU × 10 peak hours × USD 0.09/CU/hour × 30 days) + (10 CU × 14 off-peak hours × USD 0.09/CU/hour × 30 days) = USD 129.6 + USD 810 + USD 378 = USD 1,317.6
Hybrid billing	10 CU (fixed computing resources), and an elastic resource ceiling of 20 CU (billed based on actual usage)	You need to purchase only 10 CU of fixed computing resources for stable operation during off-peak hours. You can use elastic computing resources for peak hours and set a ceiling to ensure that resource usage from scaling does not exceed 20 CU.	Estimated maximum fee: Fee for management resources + Fee for computing resources = Fee for management resources + Fee for fixed computing resources + Fee for elastic computing resources = (2 CU × USD 45.94/CU/month × 1 month) + (10 CU × USD 45.94/CU/month × 1 month) + (20 CU × USD 0.09/CU/hour × 10 peak hours × 30 days) = USD 91.88 + USD 459.4 + USD 540 = USD 1,091.28

View elastic resource usage

• View elastic computing resource usage for a namespace

  For a namespace with allocated elastic computing resources, you can view the allocation and usage of both fixed and elastic resources in the following locations on the Realtime Compute for Apache Flink console: the lower-left corner, the Resource Overview area on the Overview page, and the Queue Management page.

  The Resource Overview area displays progress bars for three metrics: CU Usage, CPU Usage, and Memory Usage. Hover over a metric to view the used amount, total amount, and usage percentage for Fixed and Elastic resources.

• View deployments that use elastic computing resources

  On the O&M > Deployments page, you can filter for deployments that use elastic computing resources by using the vvpElasticity: true label.

Disable elastic computing resources

• Disable manually

  1. Ensure that no deployments are using elastic computing resources. Otherwise, you cannot disable the feature.

     You can either stop all deployments that use elastic computing resources or increase the fixed computing resources of the namespace to match the current usage. This ensures the deployments can continue running on fixed resources.

  2. In the Actions column of the target hybrid billing workspace, choose More > Disable Hybrid Billing.

  3. Click OK.

• Passive disabling (release)

  Elastic computing resources are typically released in the following two scenarios.

  • Your account has an overdue payment

    If your account has an overdue payment, elastic computing resources are disabled even if the fixed computing resources have not expired. The workspace is not released, but all deployments that use elastic computing resources are stopped. If the payment is more than 15 days overdue, the workspace reverts to the subscription method. To use hybrid billing again, you must re-enable it after you settle the payment. For more information about how overdue payments affect deployments, see Overdue payments and expiration.

  • Your fixed computing resources expire

    If a subscription workspace (fixed computing resources) is not renewed within 15 days of its expiration, the workspace is automatically released, and the elastic computing resources are also disabled.

Related documents

• To learn how to switch between subscription and pay-as-you-go billing methods, see Switch billing methods.

• View bills