Building a Data Warehouse No. 04: A Summary of logical modeling –1–

Directory

• Objective
• Principle
• Content
• Summary

Preface

The previous article described the business Modeling and domain concept modeling in the Data warehouse model design, and then came naturally to the phase of the logical Data Modeling LDM (Logical data Model), which is one of the most important aspects of modeling (that is, dimensional modeling). Logical modeling involves the design of all levels of the entire data warehouse, from DW to DM and even to OLAP. Of course, the focus of design is in the DW and DM layer.

Some areas of logic modeling are broader in scope, including the previous business topics and the design of domain conceptual models (see). This article deals only with the narrow section (the red-framed part).

content

Knowing the scope and importance of logical model design, what is the specific need to do?

• Analyze topic fields

Determine the name of the theme to be loaded into the Foundry warehouse, as well as the code keys and attribute groups for the respective topic, the entities within the theme, their capacity and update frequency, the properties of the columns of the entity, and so on

• The design of particle size model

By roughly estimating the amount of data to determine the granularity of the division, is a single granularity or multiple granularity (for example, 1 years of data is the granularity of the day, history is the size of the moon)

• Data segmentation Design

The data for an entity should be segmented in such a way that it is usually segmented by time, such as daily data in a partition.

• The establishment of the meta-data model

The metadata model can be better maintained and understood in the process of various transformations and summaries.

Principles

Logical data Model design is the core foundation of Data Warehouse project. Why this is so, because the logical data model design is principled, by satisfying these principles, can guarantee the stability of the entire data warehouse, but also make it easy for the data demand side to understand the data, processing the data is very efficient.

• Grain size

Data Warehouse different levels have different granularity, DW layer is the data is atomic granularity of data, such as transaction data atomic granularity is an order, records also include the purchase of users and merchants, DM layer data is subject to a certain dimension of the data summarized, such as the Merchant Bazaar calculates the amount of orders sold on the day.

• Sharing of

In the data warehouse, through abstraction and integration, some (dimension) information is aggregated and globally consistent so that it is shared throughout the data warehouse and can be used by any user. such as the consistency dimension.

• Historic

For the needs of business analysis, it is necessary to obtain useful information from historical information, such as evaluating customer life cycle value.

• Consistency

The logical data model must maintain a unified business definition during the design process. For example, the definition of channels, the classification of groups, etc., should be consistent throughout the enterprise. Future analytics applications use the same data, which should be refreshed in accordance with pre-agreed rules to ensure synchronization and consistency.

• Scalability

When there are new needs and changes, the structure of the logical data model can be extensible and transparent to the user.

There are, of course, other principles in which the ultimate goal of these principles is to better meet the user's use.

Summary

The logical model design is an important part of the practice dimension modeling, which will be discussed and summarized in detail from the three cores of dimension modeling.

The three cores of dimension modeling: Bus architecture, conformance dimension, and conformance facts.

Building a Data Warehouse No. 04: A Summary of logical modeling –1–