Data Waehousing

02
Data Warehouse - Basic Concepts

DW 2012/2013

Notice
! Author
" Joo Moura Pires ([email protected])

! This material can be freely used for personal or academic purposes without any previous authorization from the author, only if this notice is maintained with.

! For commercial purposes the use of any part of this material requires the previous authorization from the author.

DW Basic Concepts - 2

Bibliography
! Many examples are extracted and adapted from
" [Imhoff , 2003] - Mastering Data Warehouse Design : Relational and Dimensional Techniques, Wiley. " [Kimball, 2002] - The Data Warehouse Toolkit: The Complete Guide to Dimensional Modeling (Second Edition), from Ralph Kimball, Margy Ross, Willey

DW Basic Concepts - 3

Table of Contents
! Corporate Information Factory

! Quick overview of OLAP cube concepts

! Basics of Multidimensional Modeling

DW Basic Concepts - 4

Data Warehouse - Basic Concepts

Corporate Information Factory

DW - Basic Concepts - 5

Strategic and tactical portions of a BI environment.

[Imhoff , 2003]

Corporate Information Factory

DW Basic Concepts - 6

Corporate Information Factory Architecture

[Imhoff , 2003]

Corporate Information Factory

DW Basic Concepts - 7

CIF: Data Acquisition - (ETL)

[Imhoff , 2003]

Data acquisition is a set of processes and programs that extracts data for the data warehouse and operational data store from the operational systems. The data acquisition programs perform the cleansing as well as the integration of the data and transformation into an enterprise format. This enterprise format reects an integrated set of enterprise business rules that usually causes the data acquisition layer to be the most complex component in the CIF. In addition to programs that transform and clean up data, the data acquisition layer also includes audit and control processes and programs to ensure the integrity of the data as it enters the data warehouse or operational data store.
Corporate Information Factory DW Basic Concepts - 8

CIF: Data Delivery - (ETL)

[Imhoff , 2003]

Data delivery is the process that moves data from the data warehouse into data and oper marts. Like the data acquisition layer, it manipulates the data as it moves it. In the case of data delivery, however, the origin is the data warehouse or ODS, which already contains highquality, integrated data that conforms to the enterprise business rules.
Corporate Information Factory DW Basic Concepts - 9

CIF: Data Warehouse

[Imhoff , 2003]

a subject-oriented, integrated, time variant and non-volatile collection of data used in strategic decision making [Imnon, 1980]

Corporate Information Factory

DW Basic Concepts - 10

CIF: Operational Data Store

- It is subject oriented like a data warehouse. [Imhoff , 2003] - Its data is fully integrated like a data warehouse. - Its data is current. ! The ODS has minimal history and shows the state of the entity as close to real time as feasible. - Its data is volatile or updatable. - Its data is almost entirely detailed with a small amount of dynamic aggregation
Corporate Information Factory DW Basic Concepts - 11

CIF: Data Mart

[Imhoff , 2003]

The data in each data mart is usually tailored for a particular capability or function, such as product protability analysis, KPI analyses, customer demographic analyses, and so on.

Corporate Information Factory

DW Basic Concepts - 12

CIF: Metadata Management

Technical meta data describes the physical structures in the CIF and the detailed processes that move and transform data in the environment. Business metadata describes the data structures, data elements, business rules, and business usage of data in the CIF
[Imhoff , 2003]

Administrative metadata describes the operation of the CIF, including audit trails, performance metrics, data quality metrics, and other statistical meta data.

Corporate Information Factory

DW Basic Concepts - 13

CIF: Information feedback

[Imhoff , 2003]

Information feedback is the sharing mechanism that allows intelligence and knowledge gathered through the usage of the Corporate Information Factory to be shared with other data stores, as appropriate

Corporate Information Factory

DW Basic Concepts - 14

CIF: Information Workshop

The library component provides a directory of the resources and data available in the CIF, organized in a way that makes sense to business users. This directory is much like a library, in that there is a standard taxonomy for categorizing and ordering information components.

[Imhoff , 2003]

toolbox is the collection of reusable components (for example, analytical reports) that business users can share, in order to leverage work and analysis performed by others in the enterprise. In the workbench, metadata, data, and analysis tools are organized around business functions and tasks that supports business users in their jobs
Corporate Information Factory DW Basic Concepts - 15

Role and Purpose of the Data Warehouse

[Imhoff , 2003]

Corporate Information Factory

DW Basic Concepts - 16

The multipurpose nature of the DW

! It should be enterprise focused ! Its design should be as resilient to change as possible. ! It should be designed to load massive amounts of data in very short amounts of time. ! It should be designed for optimal data extraction processing by the data delivery programs. ! Its data should be in a format that supports any and all possible BI analyses in any and all technologies.
Corporate Information Factory DW Basic Concepts - 17

Design Pattern for the DW

! Non-redundant ! Stable
! since change is inevitable, we must be prepared to accommodate newly discovered entities or attributes as new BI capabilities and data marts are created.

! Consistent ! Flexible in Terms of the Ultimate Data Usage

Corporate Information Factory

DW Basic Concepts - 18

Design Pattern for the DW

Sta

d r a nd

a R E
+

ch a o ppr

ta a D l a c i tor s i H s e + g n a h C s e r u t c tru
Corporate Information Factory DW Basic Concepts - 19

Data Warehouse - Basic Concepts

Quick overview of OLAP cube concepts

DW - Basic Concepts - 20

Multidimensional Cube
Um negcio que products vende vrios A business sells in stores andatravs it is necessary to produtos de vrias measure the companys lojas, pretende medir o seu performance through desempenho ao longo time do tempo

Time (days) Tempo

Produtos

Products

Dollar amount Valor Sales de vendas Unit Sales Unidades vendida ...

...
Valores referentes a: Values concerning um produto a product um dia a day numa loja

Lojas Stores

Hiper-cubo Hyper-Cube

a store

Quick overview of OLAP cube concepts

DW Basic Concepts - 21

Multidimensional Cube
Time (days) Tempo Time (days)
Month Ms

Semana Week
Produtos Products

Tipos de produtos Products Type Marca de produtos Products Brand

Regio N Region N

dia day
Lojas Stores

Matriz Sparce esparsa Matrix

Medidas referentes a: Values concerning um produto a product a day um dia a storeloja numa

Quick overview of OLAP cube concepts

DW Basic Concepts - 22

Basic operation: Slice

Slice ::subconjunto Slice a subset of dos dados multidimensional data Multidimensionais. Um slice definido atravs Slice : a slice is defined by da seleco valores especficos selectingde specific values of dimensions para atributosattributes das dimenses

Quick overview of OLAP cube concepts

DW Basic Concepts - 23

Basic operation: Aggregation

Tempo

l = l2 , t = t1 , p

f ( l, p, t )

Produtos

l{l2 , l3 ,l5}, t = t1 , p
Lojas

f (l, p, t )

l, t = t1 , p MarcaX

f (l, p, t )

Regio 1

l, t = t1 , p MarcaY

f (l, p, t )

Quick overview of OLAP cube concepts

DW Basic Concepts - 24

Data Warehouse - Basic Concepts

Basics of Multidimensional Modeling

DW - Basic Concepts - 25

Multidimensional Cube
! A Data Modeling approach with the purpose of addressing the following aspects: ! The resulting data models should be understandable by the analytical users: ! Simple. ! Using terms from the domain and appropriate for data analysis. ! Provides a framework for efficient querying ! Provides the basics for generic software development where the users can navigate in large data sets in an intuitive way

Basics of Multidimensional Modeling

DW Basic Concepts - 26

Star schema
! Fact table
! Big and central table. The only table with many joins connecting with the others tables

! Many Dimension Tables

! With only one join connecting to the fact table

Asymmetric Model

Product Time
time_key day_of week month quarter year

Sales
time_key product_key store_key value units cost

product_key description brand category

Dimension

Store
loja_key name address type

Dimension

Fact Table
Basics of Multidimensional Modeling

Dimension

DW Basic Concepts - 27

Fact Tables
! Numerical measures of process.
! Continuos values (or represented as continuos values). ! Additive (may be correctly added by any dimension). ! Semi-additive (may be correctly added by some dimension but not on other dimensions). ! Non-additive (cannot be added but some other aggregation operators are allowed)

! The goal is to summarize the information presented in fact tables. ! The granularity of a fact table is defined by a sub-set of dimensions that index it.
! Ex: sales per day, store and product.

! Fact tables are, in general, sparse

! Ex: If a product is not sold on a day, in a store then there is no correspondent record on the fact table.

Basics of Multidimensional Modeling

DW Basic Concepts - 28

Dimension Tables
! Tables with simple primary keys that are related to fact tables. ! The most interesting attributes the ones with textual descriptions.
! They are used to define constraints over the data that will be analyzed. ! They are used to group the aggregations made over the fact table measures. They will be the headers columns

Brand Marca M-1 M-2 M-3 M-4

Dollar sold Valoramount Vendido

780 1044 213 95

Sold Vendidas Units Unidades 263 509 444 39

Basics of Multidimensional Modeling

DW Basic Concepts - 29

Typical result
! Data for the first quarter for all stores by brand

Brand Marca M-1 M-2 M-3 M-4

Dollar amount sold Valor Vendido

780 1044 213 95

Sold Units Unidades Vendidas 263 509 444 39 Metrics

Distinct values for the selected attribute Textual Attribute of a Dimension

Basics of Multidimensional Modeling DW Basic Concepts - 30

Querying a Star Schema

Dimension Product Time
time_key day_of week month quarter year

Sales
time_key product_key store_key value units cost

product_key description brand category

Store
loja_key name address type

Dimension

Fact Table

Dimension

Basics of Multidimensional Modeling

DW Basic Concepts - 31

Typical SQL query for OLAP

Selecting the columns

select p.brand, sum(f.value), sum(f.units) from sales f, product p, time t where f.product_key = p.product_key ! and f.time_key = t.time_key ! and f.quarter = Q1 1996 group by p.brand order by p.brand

aliases Join constraint Join constraint Application constraint

Grouping Sorting

Basics of Multidimensional Modeling

DW Basic Concepts - 32

Processing the SQL query for OLAP

! First, the application constraints are processed for each dimension
! Ex: Month = Mars; Year = 1997; Type of store = Hyper; Region = ..; ...

! Each dimension produces a set of candidate keys:

! Ex: Time: All time_key for which Month = Mars; Year = 1997;

! All the candidate keys are concatenated (Cartesian Product) to get the keys to be searched in the fact tables. ! All the hits on the fact table are grouped and aggregated.

Basics of Multidimensional Modeling

DW Basic Concepts - 33

Browsing the Dimension Tables

! Dimension Browsing - is the user activity where the user explore the data in the dimensions with the purpose of defining constraints over the dimensions attributes and to select the level and type of intended summarization for the OLAP answers. ! Generic and convenient mechanism used by the user to specify the Queries.
! SIMPLICITY ! PERFORMANCE

Basics of Multidimensional Modeling

DW Basic Concepts - 34

Browsing the Dimension Tables

Dimenso: dim1 (ex: produto) Atributo: Marca Tipo
Telemvel

Nome

Restrio: Alcatel

Nokia
Valores Distintos: Alcatel Ericson Coca-Cola Nokia Motorola Nestle Telemvel Televiso ...
Easy .. .. 3610 ...

Basics of Multidimensional Modeling

DW Basic Concepts - 35

Drill Down e Drill Up

Department Dep. D-1 D-2 D-3 D-4 Sales Amount Valor Vendido 780 1044 213 95 Sales Units Unidades Vendidas 263 509 444 39

Desagregou departamento marca Drill down to departmentpor and Brand

Department Dep. D-1 D-1 D-2 ...

Brand Marca M-1 M-2 M-5

Sales Amount Valor Vendido 300 480 ... .

Sales Units Unidades Vendidas 160 103 .. ..

DW Basic Concepts - 36

Basics of Multidimensional Modeling

Drill Down e Drill Up

! Drill down is just to add some new header columns to the result table, which is a dimension attribute

! Drill-Up is the reverse operations

Basics of Multidimensional Modeling

DW Basic Concepts - 37

From a rowset to an analytical view

Basics of Multidimensional Modeling

DW Basic Concepts - 38

Classical OLAP view

Basics of Multidimensional Modeling

DW Basic Concepts - 39

Inefficient OLAP view

Basics of Multidimensional Modeling

DW Basic Concepts - 40

What about Partial Totals?

Sum of Sales Diviso Tipo_Prod ACCESSORY AUDIOTAPE - DIV Trimestre PROD T1 T2 T3 T4 Grand Total C1-AUDIOTAPE 12128.13 11932.07 7016.2 8354.66 39431.06 C1-CHROMECAS 1311.39 1258.68 688 936.42 4194.49 C1-METALCAS 8335.54 8258.47 4836.6 5502.66 26933.27 C1-STNDCAS 2481.19 2414.93 1491.6 1915.58 8303.3 AUDIOTAPE Total 24256.25 23864.15 14032.4 16709.32 78862.12 VIDEOTAPE C2-8MMVIDEO 9657.51 10222.88 5437.3 6392.68 31710.37 C2-HI8VIDEO 10739.28 10600.47 5778.5 7140.94 34259.19 C2-STNDVHSVIDEO 6396.91 6472.93 4057.8 5594.56 22522.2 VIDEOTAPE Total 26793.7 27296.28 15273.6 19128.18 88491.76 ACCESSORY - DIV Total 51049.95 51160.43 29306 35837.5 167353.88 AUDIO - AUDIO DIV - COMP A2-AMPLIFIER 108876.35 99776.02 54242.3 62432.28 325326.95 A2-CASDECK 20434.01 17162.82 8551.8 11360.34 57508.97 A2-CDPLAYER 148301.35 121497.44 59753.6 78906.74 408459.13 A2-RECEIVER 86468.12 90890.41 50763.2 60066.96 288188.69 A2-TUNER 28830.88 26136.36 13724.4 16752.34 85443.98 AUDIO - COMP Total 392910.71 355463.05 187035.3 229518.66 1164927.72 PORT-AUDIO A1-PORTCAS 21857.27 22936.96 11720.8 16388.68 72903.71 A1-PORTCD 37139.63 30166.12 13803.3 18002.58 99111.63 A1-PORTST 30241.77 31871.52 17446.2 21478 101037.49 PORT-AUDIO Total 89238.67 84974.6 42970.3 55869.26 273052.83 AUDIO - DIV Total 482149.38 440437.65 230005.6 285387.92 1437980.55 VIDEO - DIV CAMCORDER B3-8MMCMCDR 127708.61 122016.17 66015.4 82212.2 397952.38 B3-HI8CMCDR 90308.93 93434.34 45232.3 56331.22 285306.79 B3-VHSCMCDR 154074.17 147218.21 81591.7 97779.32 480663.4 CAMCORDER Total 372091.71 362668.72 192839.4 236322.74 1163922.57 TV B1-BWTV 11426.3 11984.54 6675.7 8512.42 38598.96 B1-COLORTV 23693.66 19846.51 10117.1 12954.52 66611.79 B1-PORTTV 15914.94 14511.87 7265.9 7864.24 45556.95 TV Total 51034.9 46342.92 24058.7 29331.18 150767.7 VCR B2-STNDVCR 21199.71 19816.63 11910.1 13569.5 66495.94 B2-STRVCR 37818.57 39045.7 19096.7 23015.96 118976.93 B2-TOTALPROD 595283.24 575747.89 325688.3 404670.1 1901389.53 VCR Total 654301.52 634610.22 356695.1 441255.56 2086862.4 VIDEO - DIV Total 1077428.13 1043621.86 573593.2 706909.48 3401552.67 Grand Total 1610627.46 1535219.94 832904.8 1028134.9 5006887.1

Basics of Multidimensional Modeling

DW Basic Concepts - 41

Data Warehouse - Basic Concepts

Further Reading and Summary

DW - Basic Concepts - 42

Further Reading and Summary

! Readings ! (Kimball - The Data Warehouse toolkit, 2002) - pag 16 to 27.

! What you should know: ! Understand the Corporate Information Model (CIF): The different roles for the DW, the ODS and the Data Marts (specially the OLAP data marts). The fundamental aspect of feedback from the knowledge and information gathered at DSS systems into the architecture (operational systems and the DW) ! Understand the fundamental differences between OLTP and the analytical activities developed on the DW or on the Data Marts: data, access, users ...

DW Basic Concepts - 43