When using DataWorks for data modeling, data warehouse architects and modeling teams can design data layers, business categories, data domains, business processes, data marts, and subject areas on the warehouse planning page. After the design is complete, model designers can use these elements to manage models by layer and domain.
Overview
Before starting data modeling, data warehouse architects must collaborate with data development and model design teams to investigate business requirements to clarify the overall data structure. Complete the following core designs during the data warehouse planning phase to effectively manage models by layer and domain.
Business categories: a vertical division for complex business scenarios, such as e-commerce or finance.
Data domains and business processes: an abstraction of key business links.
Data marts and Subject areas: data aggregation for specific business scenarios.
Data warehouse layers: a layered logic that includes the Operational Data Store (ODS), Dimension (DIM), Data Warehouse Detail (DWD), Data Warehouse Summary (DWS), and Application Data Service (ADS) layers.
Business planning
Business categories, data domains, and data marts form a business-driven management framework. By dividing data into specific domains (business categories), defining core business activities (data domains), and organizing scenario-based data services (data marts), you can create a closed-loop value chain from data production to consumption. Data warehouse layering is a technology-driven data processing pipeline that implements this framework and refines raw data into a service-ready state.
Business categories: The highest-level division of business areas, such as e-commerce, finance, and retail.
Data domains: A high-level data classification standard created by abstracting, refining, and combining business processes. Data domains divide business data based on dimensions such as business type, data source, and data usage. A data domain can belong to multiple business categories. For example, the transaction domain can serve transaction scenarios in both e-commerce and finance.
Business processes: Specific business activities within a data domain. For example, a transaction domain can contain business processes like placing an order and making a payment. A data domain can have multiple business processes.
Data marts: Data outlets for specific business scenarios, such as an operations platform mart.
Subject Areas: A division of data marts based on analysis perspectives, such as product analysis or user behavior analysis. A data mart can have multiple subject areas.
Technical planning
DataWorks provides a default five-layer data warehouse architecture (ODS, DIM, DWD, DWS, and ADS) that aligns with industry standards and meets the needs of most data warehouse development projects. You can also customize the layers in Data Layer:
Layer group | Data warehouse layer | Abbreviation | Main function | Supported model types | Supported metric types |
Data Import Layer | Operational data store | ODS | Receives and processes raw data. The structure is consistent with the source system. | Source table | - |
Common Layer | Dimension | DIM | Builds consistent enterprise-level dimension tables. | Dimension table and dimension | Atomic metrics |
Data warehouse detail | DWD | Creates fact tables with detailed data, typically as wide tables. | Detail table | Atomic metrics | |
Data warehouse summary | DWS | Constructs summary metrics at a public granularity. | Aggregate table | Atomic metrics, composite metrics, and derived metrics | |
Application Layer | Application data service | ADS | Stores personalized statistical metrics. | Application table,dimension table, and dimension | Composite metrics and derived metrics |
Data warehouse layering is an important technical management tool for data warehouse planning. It provides a vertical structure for the entire data warehouse that spans all business categories, data domains, and data marts. Each layer maintains mappings to the relevant business categories and data domains or data marts.
Implementation recommendations
Plan and design your own data warehouse
If you want to meet specific enterprise requirements, you can customize your data warehouse planning. We recommend that you first clarify your business objectives (for example, a "member growth analysis" requirement belongs to the member domain) and then design the technical solution (for example, design a DWD member detail table).
Plan business categories, data domains, and data marts.
Design table storage levels based on the five-layer architecture.
Use a checker to standardize the naming conventions for each layer.
For complex enterprises, enable modeling space to reuse architectures.