Model Call Girl in Tilak Nagar Delhi reach out to us at 🔝9953056974🔝
Building Data Warehouses with Relational and Multidimensional Models
1. Building Data WareHouse by
Inmon
Chapter 13: The Relational and the Multidimensional
Model as a Basis for Database Design
http://it-slideshares.blogspot.com/
2. Contents
The Relational Model
The Multidimensional Model
Snowflake Structures
Differences between the Models
Independent Data Marts
Building Independent Data Marts
Summary
3. The Relational Model
Organize data into a table
Normalization of data implies that the database design
has caused the data to be broken down into a very low
level of granularity
4. The Multidimensional Model
Sometimes
called the star join approach
Components:
◦ A fact table is a structure that contains many occurrences of
data.
◦ Dimensions table, which describe one important aspect of the
fact table
5. Snowflake Structures
More than one fact table can be combined in a database
design to create a composite structure called a
snowflake structure
Advantage of the multidimensional design is its
efficiency of access
6. Differences between the Models
The Roots of the Differences
◦ The relational model is shaped from a data
model.
◦ A star join is shaped from user requirements.
7. Differences between the Models
Reshaping Relational Data
◦ The base data in the relational model can be
shaped and reshaped in as many ways as
desired
8. Differences between the Models
Indirect Access and Direct Access of
Data
◦ The relational model is good for indirect
access of data
◦ The multidimensional model is good for
direct access of data
9. Differences between the Models
Servicing Future Unknown Needs
◦ The granular data in the relational model is
used to service unknown future needs for
information
10. Differences between the Models
Servicing the Need to Change Gracefully
◦ Another advantage of the relational model as
a basis for the data warehouse—the ability to
change gracefully
◦ The impact of change is minimal
11. Differences between the Models
Servicing
the Need to
Change Gracefully
◦ The relational model forms an
ideal basis for the data
warehouse, while the star join
forms the ideal basis for the
data mart.
12. Independent Data Marts
Data Marts
◦ A data mart is a data structure that is
dedicated to serving the analytical needs of
one group of people
◦ The independent data mart is a data mart that
is built directly from the legacy applications
13. Independent Data Marts
Data Marts
◦ A dependent data mart is one that is built from
data coming from the data warehouse
◦ The dependent data mart requires multiple
users to pool their information needs for the
creation of the data warehouse
14. Building Independent Data Marts
Problem with building independent data marts
◦ Do not provide a platform for reusability
◦ Do not provide a basis for reconciliation of data
◦ Do not provide a basis for a single set of legacy interface
programs
◦ Do require that every independent data mart create its own
pool of detailed data, which is, unfortunately, massively
redundant with the pools of detailed data created by other
independent data marts
15. Building Independent Data Marts
With dependent data marts, all problems of
independent data marts are solved
16. Summary
Basic models that are used for database design for the
data warehouse: the relational model and the
multidimensional (star join) model
The relational model is ideal for serving indirect access
to the data warehouse, while the multidimensional
model is ideal for serving the needs of the direct use of
the data warehouse.
Dependent data marts that take data from a data
warehouse do not have to have the same set of
architectural problems (like independent data marts).
http://it-slideshares.blogspot.com/
Notes de l'éditeur
There are two basic models for database design that are widely considered—the relational model and the multidimensional model. The relational model is widely considered to be the “Inmon” approach, while the multidimensional model is considered to be the “Kimball” approach to design for the data warehouse.
In a snowflake structure, different fact tables are connected by means of sharing one or more common dimensions. Sometimes these shared dimensions are called conformed dimensions.
The merging of relational tables to create a new relational table is easy for several reasons: ■■ Data is stored at the most granular, normalized level. ■■ Relationships between relational tables are already identified and have a key-foreign key manifestation. ■■ New tables can contain new summaries, new selection criteria for those summaries, and aggregations of the base data found in the relational table.
No guarantee that the star join that is optimal for one group of users will contain the data needed for another group of users
The relational model is designed to be used in an indirect fashion. This means that the direct users of the data warehouse data access data that comes from the relational model, not data in the relational model itself. When it comes time for change, the impact is minimal because the different users of the data warehouse are accessing different databases.
The dependent data mart requires multiple users to pool their information needs for the creation of the data warehouse. In other words, the dependent data mart requires advance planning, a long-term perspective, global analysis, and cooperation and coordination of the definition of requirements among different departments of an organization.
Independent data marts represent a short-term, limited scope solution where it is not necessary to look at the global, long-term picture. Dependent data marts, on the other hand, require a long-term and a global perspective. But independent data marts do not provide a firm foundation for corporate information, while dependent data marts indeed do provide a sound long-term foundation for information decisions.