Grasping the logical structure of Dataphin's core concepts is crucial for providing insights into project planning and modeling. This topic elucidates the logical structure and the essence of the core concepts within Dataphin.
The logical structure of Dataphin's core concepts is depicted in the figure below.
The architecture of Dataphin comprises the following levels:
Business model layer and compute engine layer: The business model layer redefines and organizes data from a business perspective by classifying and tagging it, while the compute engine layer performs the actual data computation and storage.
The business model layer is segmented into business segments based on different business forms, with each form corresponding to a distinct segment.
Within a given business form (segment), business entities (dimensions and business processes) are categorized into various data domains based on the specific business context.
Logical tables are meticulously crafted (including logical dimension tables and logical fact tables) based on dimensions and business processes, and metrics are defined (including atomic metrics, business filters, statistical periods, statistic granularity, statistical timeliness, and derived metrics).
In Dataphin version 3.3, dimensions have been renamed to business objects, business processes to business activities, data domains to subject areas, and business segments to data segments.
For a comprehensive explanation of the core concepts at each layer, refer to the table below.
Core Concept | Brief Meaning |
Data Segment | Data segments define various namespaces of the data warehouse and are a system-level conceptual object. When the business meaning of data varies greatly, you can create different data segments to allow each member to independently manage different businesses. The subsequent construction of the data warehouse will be divided according to data segments. |
Subject Area | Data domain, also known as subject area, is the boundary of a subject determined after analyzing a particular subject. For example, product domain, transaction domain, member domain, etc. |
Project | A project is a division in physical space that facilitates the isolated management of physical resources and developers during the construction of the data mid-end. |
Business Entity | The perspective from which people observe things, referring to a viewpoint, is the condition and concept for determining the multi-faceted, multi-angled, and multi-layered nature of things. |
Business Activity | Business process, also known as business activity event, is usually an indivisible event. It is an activity or the result of an activity conducted by one or more business objects at a certain time or period to achieve a certain purpose. |
Logical Dimension Table | A logical table formed by enriching the attribute information of dimensions. Through logical dimension tables, you can design and process detailed data of public objects to extract detailed data of objects in business. |
Logical Fact Table | Used to describe detailed information of business processes. By creating logical fact tables, you can design and process detailed data of public transactions to extract detailed data of transactions in business. |
Atomic Metric | An abstraction of the statistical caliber and specific algorithm of a metric. For example, payment amount. |
Business Filter | The business scope of statistics, used to filter records that meet business rules (similar to the conditions after Where in SQL, excluding time intervals). |
Statistical Period | Defines the time span of the source data for derived metrics. For example, the last 1 day, the last 30 days, etc. (similar to the time conditions after Where in SQL). |
Statistic Granularity | The object or perspective of statistical analysis, used to delineate the statistical scope of data. You can also understand it as the grouping condition during aggregation operations (similar to the object of Group By in SQL). |
Statistical Timeliness | The calculation frequency of derived metrics, which is the time interval at which derived metrics are produced. |
Derived Metric | Based on atomic metrics, period, and dimensions, delineates the business statistical scope and analyzes to obtain the value of business statistical metrics. |
Logical Aggregate Table | The table to which derived metrics belong is the logical aggregate table. |
Physical Table | The table in the compute engine, created through DDL. |
Materialized Table | The physical table that stores the real data of the logical table. |