Scheduled tasks require resource groups to run. Based on whether they consume CUs from a resource group, tasks are categorized as scheduling tasks and compute tasks. Scheduling tasks are only triggered and monitored by DataWorks and do not consume compute resources (CUs). Compute tasks execute code within DataWorks and consume compute resources. Configuring resource properties for tasks in the scheduling configuration ensures stable operation and optimizes costs.
Key concepts
Tasks are divided into two types based on whether they consume DataWorks Compute Units (CUs) during execution:
|
Task type |
Core responsibility |
CU consumption |
Task list |
|
scheduling task |
Triggers and monitors computations in external compute engines, such as MaxCompute and Hologres. |
Does not consume |
|
|
compute task |
Executes code directly within a DataWorks resource group to perform computations. |
Consumes |
-
This topic focuses on CU configurations for
Serverlessresource groups. If you use dedicated or subscription resource groups, you only need to specify the resource group and do not need to configure CUs. -
This topic does not cover the resource consumption or billing of third-party engines called by tasks, such as MaxCompute.
Procedure
You must configure scheduling resources for all scheduled tasks. The configuration process varies slightly depending on the task type and resource group type.
Step 1: Specify a scheduling resource group
This is the baseline configuration for all scheduled tasks.
-
In the task editor, find the Scheduling Settings panel on the right.
-
In the Resource Group section, select a scheduling resource group from the drop-down list.
ImportantWhen selecting a resource group, ensure that it can connect to the data sources and compute resources that the task requires. Otherwise, the task might fail. For more information about network connectivity, see: Network Connectivity Configuration.
After you make a selection, a message indicates that the current task is a scheduling task and does not require CUs.
Step 2: Configure compute resources (for compute tasks only)
If your task is a compute task and you selected a Serverless resource group in Step 1, you must perform this step.
Tip: You can skip this step for scheduling tasks or non-Serverless resource groups.
-
A CUs for Scheduling configuration option automatically appears below the Resource Group for Scheduling setting.
-
Set the number of CUs required for runtime (for example: 0.25 CUs, 1 CU, or 4 CUs) based on the task's compute complexity and performance needs. An appropriate CU configuration ensures stable operation and helps optimize costs.
Step 3: Configure an image
For task types such as PyODPS and Shell, you can also define the task's runtime environment by configuring an Image. For more information about image management, see: Custom Images.
The configuration form includes the following fields: scheduling resource group (required, for example, select dataworks_default_resource_group), scheduling CU (for example, enter 0.25), and image (select one from the drop-down list).
Set the default scheduling resource group
When you create a new task, the scheduling resource group specified in Scheduling Settings becomes the default resource property. To adjust this, click the 
icon in the lower-left corner, navigate to , and modify the default scheduling resource group.
Billing
Running scheduled tasks on a Serverless resource group incurs two types of fees:
-
task scheduling fees: These fees are based on the number of successful task runs per day. For more information, see Scheduling instance fees.
-
resource usage fees: These fees apply only when a compute task runs on a pay-as-you-go Serverless resource group. The fees are calculated based on the consumed CUs and the task's runtime. These fees do not include charges from third-party engines called by the task. For more information, see Serverless resource group billing.
For data integration tasks, resource consumption depends on the consumed CUs you specify for the resource group, not the scheduling CU setting.