How DataWorks’ New Serverless Resource Groups Cut Costs and Boost Flexibility

Background Introduction

DataWorks resource groups provide compute resources for various DataWorks modules and are a paid service.

Resource groups are a fundamental component of DataWorks, prerequisite for normal use.

They directly affect functionality, efficiency, and stability.

Previously, resource groups were divided into public integration, public scheduling, dedicated integration, dedicated scheduling, and dedicated data service, leading to many concepts, complex billing, and inflexible usage; the new generic resource group was created to address these issues.

The new resource group can be used by all DataWorks functions with clear and simple billing.

The Core Feature of the New Resource Group – Serverless

Serverless is an architectural concept where applications are packaged as functions and run without the user managing servers; resources are allocated, scaled, and billed on demand.

Scheduling tasks: Fully serverless with pay‑per‑use.

Data integration, computation, services: Pay‑per‑use, with optional annual/monthly packages.

Key Advantages of the New Resource Group

Generic: Any DataWorks capability can use it.

Flexible payment: Supports both pay‑as‑you‑go and subscription.

Elastic scaling without impact: Scaling does not affect running tasks.

Scheduled scaling: Users can set scaling plans based on workload patterns.

Zero waste: Pay only for used resources, minimum granularity 1CU.

High isolation and security: Resources are dedicated to the user with user‑controlled network.

Low‑Cost Pay‑As‑You‑Go

The new resource group abandons fixed‑spec billing and charges based on actual usage.

Billing Example

A user synchronizes MySQL data to MaxCompute every day at midnight, launching 20 sync tasks, each requiring 0.5 CU for 1 hour.

Daily cost = 20 × 0.5 × 1 CU × price per CU = 10 CNY.

Previously the smallest required ECS instance (4c8g) costs 492.5 CNY per month, about 16.4 CNY per day.

Dynamic Scaling (Coming Soon)

Users no longer need capacity planning; resources are provisioned on demand.

Resources can be increased or decreased during task execution based on traffic.

Scaling is smooth, without interrupting business requests.

Comparison of Old and New Resource Groups

New Resource Group Billing

Billing is divided into prepaid (annual/monthly) and postpaid (pay‑as‑you‑go) resource groups, which cannot be converted.

Minimum Configuration Specs for Each Task Type

Data Compute Charges in the New Resource Group

DataWorks supports various node tasks (e.g., PyODPS, EMR Hive). Some tasks are dispatched to external engines and incur engine fees; tasks executed on the resource group are charged by DataWorks.

Tasks sent to compute engines are billed by the respective engine.

Tasks run on the new resource group are billed by DataWorks.

Running data‑compute tasks in DataWorks (e.g., in DataStudio, data quality, data analysis, or operation center) will generate compute fees.

For more details, refer to the comparison and usage documentation.