Data Warehouse Architecture and Models

Data Warehouse Architecture and Models

Lesson 4

Data Warehouse Architecture and Models

 Explain the features of each type of data by

examining where and why it is used

 List the data models that may already exist in a company and describe where they may be

useful to the warehouse model

 Explain the two common data warehouse

Warehouse Architectures

 Enterprise-wide solution

 Data mart solution

 Combined solution

Enterprise Data Warehouse Solution

Data Mart Solution

 Independent

 Use a consistent approach

 Avoid disjointed development

 Consider the big picture

 Dependent

 Localization

 Subsets of summary data

 Oracle Data Mart Suite for a pre-configured solution

 Funded departmentally

Independent Dependent

Create a Project Team

 Staff with experts

Identify the Data Requirements

 Successful warehouses provide the right information

 Analyze business users’ needs

 Interview users

 Examine data needed

 Ascertain data availability

 Determine data frequency

 Decide the refresh cycle

Types of Warehouse Data

 Fact data - Measures

 Dimension data - Query drivers

 Summary data - Pre-calculated data

Fact Data and Tables

 Many fact tables in the warehouse

 The bulk of the warehouse data

 Measures (units sold, sales figures, calls)

 Millions of rows

 Multi-part primary keys

 Summaries

 Normalized data

Vertical Partitioning

 Vertical partitioning by column

Col A Col B Col C Col D

• Oracle servers support both types of partitioning

Dimension Data

Query path

Customer Location

Product Time

Sales

• Provides query criteria

• Links to fact tables with keys

Dimension Tables

 Are determined by user requirements

 Vary in number

 May contain hierarchies

 May share fact tables

Time Dimension

 Important required dimension

 Contains special dates

 Provides flexible and accurate analysis by time

Customer Office Office

Product Time

Summary Tables

 Provide immediate answers to a query

 Improve query performance

 Requirements change and need managing

SALES FACTS Sales$ Region Month

Jan 97 East 10,000 Feb 97 South 40,000 Mar 97 West 17,000

SALES BY MONTH Month Tot_Sales Jan 97 51,000 Feb 97 40,000 Mar 97 17,000

 Define the model

 Use graphical modeling tools

 Use a tool capable of prototyping

 Proof of concept essential

 New design techniques

The Enterprise Model

 Defines an overall scope

 A good start point for analysis

 An information framework for the warehouse

The Corporate Data Model

 Current operational data structures

 Source of mapping rules for the warehouse

Marketing

A Typical Modeling Approach

 1 Analyze the subject area.

 2 Gather the requirements.

 3 Develop the models.

 4 Map the entities, attributes, and

relationships.

 5 Re-engineer the source data.

 6 Design the database.

 7 Integrate the model into the

warehouse architecture

repository.

 8 Review with client and revise.

iterative approach.

Sales Fact Table Item_id

Store_id Sales_dollars Sales_units

Store Table Store_id District_id

Store Table Store_id District_id

Item Table

Item_id

Item Table

Item_id Dept_id

Time Table Week_id

Time Table Week_id Period_id

Totsales Summary Month_id

Store_id Item_id Total_dollars

Totsales Summary Month_id

Store_id Item_id Total_dollars

Snowflake Model

Sales Fact Table Item_id

Store_id Sales_dollars Sales_units

Sales Fact Table Item_id

Store_id Sales_dollars Sales_units

Store Table Store_id Store_desc District_id

Store Table Store_id Store_desc District_id

Item Table Item_id Item_desc Dept_id

Item Table Item_id Item_desc Dept_id

Totsales Month_id Store_id Item_id Total_dollars

District Table District_id District_desc

District Table District_id District_desc

Dept Table Dept_id Dept_desc

Dept Table Dept_id Dept_desc

Mgr Table Dept_id Mgr_id

Mgr Table Dept_id Mgr_id

Snowflake Model

 A normalized star schema

 Easier to model requirements

 Flexible dimension structures

 Readily maps to existing data

 Used directly by tools

 Database servers

 Star queries

 Star joins

 VLDB support

Which Model Do you Use?

 Design for simplicity

 Design for relevance

 Simple star model

 Flexible snowflake model