Data Warehouse Modeling

Data Warehouse Modeling

Thijs Kupers

Vivek Jonnaganti

Agenda• Introduction

• Data Warehousing Concepts

• OLAP

• Dimension Modeling

• Conceptual Modeling

• Indexing

• Conclusion

Introduction

The Evolution• 1960 - DSS processing using Fortron or COBOL

• 1970 - DBMS systems and the advent of DASD

• 1975 - OLTP systems facilitating faster access to data

• 1980 - PC/4GL technology and the advent of MIS

• 1985 - OLAP systems and separation of analytical processing from transactional processing

• 1994 - Architectured environments with integrated OLAP engines and tools

What is a Data Warehouse?• A copy of transaction data specifically structured to Query

and Analysis (Ralph Kimball, 1996)

• A collection of integrated, subject oriented databases designed to support the DSS function where each unit of data is relevant at some moment of time (Bill Inmon, 1991)

• The data characteristics of a Data Warehouse are;• Subject-oriented

• Time-variant

• Non-volatile

• Integrated

What is a Data Warehouse? (cont’d)

• A single, complete and consistent store of data obtained from a variety of different sources made available to end users, in what they can understand and use in a business context (Barry Devlin 1992)

• A process of transforming data into information and making it available to users in a timely enough manner to make a difference (Forrester Research 1996)

Data Warehouse Goals/Characteristics• It must make an organization’s information easily accessible

(slicing and dicing)

• It must present the organization’s information consistently

• It must be adaptive and resilient to change

• It must be a secure bastion that protects our information assets

• It must serve as the foundation for improved decision making

• The business community must accept the DW, if it is to be deemed successful

Data Warehouse Applications• Retail Industry

• Forecasting, Market research, Merchandising etc.

• Manufacturing and distribution• Sales history/trends, Market demand projects etc.

• Banks• Spot market trends, Marketing, Credit cards etc.

• Insurance Companies• Property and casualty fraud etc.

• Health Care Providers• Fraud detection, Patient matching etc.

Data Warehouse Applications• Government Agencies

• Auditing tax records, information sharing across different agencies etc.

• Internet Companies• Analyzing shopping behavior, CRM etc.

• Telecommunications• Telemarketing, Product development etc.

• Sports• Analyzing strategies, Winning player combinations etc.

Data Warehouse Sizes• Terabyte (10^12) - Walmart (24 TB)

• Petabyte (10^15) - Geographic Information Systems

• Exabyte (10^18) - National Medical Association

• Zettabyte (10^21) - Weather Images

• Zottabyte (10^24) - Intelligence Agency (Video)

Data Warehousing Concepts

Data Warehouse (OLAP) and OLTP

CharacteristicsOn-Line Transaction Processing (OLTP) Data Warehouse

Data Content Current values Historical data, summarized data, calculated data

Data Organization Application by application Subject areas across enterprise

Nature of Data Dynamic Static until refreshed, based on frequency

Data Manipulation Updated on a field-by-field basis

Accessed & manipulated usually no direct update

UsageHighly structured, repetitive processing (Clerical User)

Highly structured, analytical processing (Knowledge User)

Response TimeCritical (Sub-Second to several seconds) Several seconds to minutes

Updates vs. Reports

Real-time Updates, Batch Reporting

Batch Updates,Real-time Reporting

Data Warehouse Architecture

Enterprise

Data

Warehouse

Enterprise

Data

Warehouse Data MartData Mart

Data MartData Mart

ExecutionSystems

• CRM• ERP• Legacy• e-Commerce

ExecutionSystems

• CRM• ERP• Legacy• e-Commerce

•Reporting Tools

•OLAP Tools

•Ad Hoc Query Tools

•Data Mining Tools

•Reporting Tools

•OLAP Tools

•Ad Hoc Query Tools

•Data Mining Tools

•External Data

• Purchased Market Data• Spreadsheets

•External Data

• Purchased Market Data• Spreadsheets

•Oracle•SQL Server•Teradata•DB2

•Custom Tools•HTML Reports•Cognos•Business Objects•MicroStrategy•Oracle Discoverer•Brio•Data Mining Tools•Portals

Data and Metadata Repository Layer

•Informatica PowerMart•Ab Initio•Data Stage•Oracle Warehouse Builder•Custom programs•SQL scripts

Extract, Transformation, and Load (ETL)

Layer

• Cleanse Data• Filter Records• Standardize Values• Decode Values• Apply Business Rules• Householding• Dedupe Records• Merge Records

Extract, Transformation, and Load (ETL)

Layer

• Cleanse Data• Filter Records• Standardize Values• Decode Values• Apply Business Rules• Householding• Dedupe Records• Merge Records

Presentation Layer

ETL LayerOperational

Source Systems

Technologies:

Metadata RepositoryMetadata Repository

ODSODS

•PeopleSoft•SAP•Siebel•Oracle Applications•Custom Systems

Data MartData Mart

Data Warehouse Structure

DepartmentallyStructured

IndividuallyStructured

Data WarehouseData WarehouseOrganizationallyStructured

Data

Information

Highly Summarized

Lightly Summarized

Atomic/Detailed

Data Warehouse Architecture DriversThe requirements that drive the DW architecture are;

• Granularity of data

• Data retention and timeliness

• Reporting capability

• Availability

• Scalability

Data Mart Centric

Data Marts

Data Sources

Data Warehouse

Data Mart Centric

If you end up creating multiple warehouses, integrating them is a problem

Data Warehouse Centric

Data Marts

Data Sources

Data Warehouse

OLAP

OLAP: 3 Tier DSS

Data Warehouse

Database Layer

Store atomic data in industry standard Data Warehouse.

OLAP Engine

Application Logic Layer

Generate SQL execution plans in the OLAP engine to obtain OLAP functionality.

Decision Support Client

Presentation Layer

Obtain multi-dimensional reports from the DSS Client.

OLAP Servers• Support multidimensional OLAP queries

• Characterized by how the underlying data is stored

• Multidimensional OLAP (MOLAP) Servers• Data stored in array based structures e.g. Hyperion

Essbase

• Relational OLAP (ROLAP) Servers• Data stored in relational tables e.g. Microstrategy, IBM

Informix

• Hybrid OLAP (HOLAP) Servers• Data distributed between relational and specialized

storage e.g. Cognos, Microsoft Analysis Services

OLAP Operations• Rollup; summarize operations

• E.g. given sales data, summarize sales for last year by product category and region

• Drill down; get more details• E.g. given summarized sales as above, find breakup of

sales within each region

• Slice and dice; select and project• Sales of soft-drinks in Gothenburg over the last quarter

• Pivot; change the view of data

Strengths of OLAP• It is a powerful visualization tool

• It provides fast, interactive response times

• It is good for analyzing time series

• It can be useful to find some clusters and outliners

• Many vendors offer OLAP tools

Dimensional Modeling

What is Dimensional Modeling?• Logical design technique that seeks to present the

data in a standard, intuitive framework that allows for high-performance access.

• Adheres to a discipline that uses the relational model with some important restrictions.

• Composed of one table with a multi-part key, called the fact table, and a set of smaller tables called dimension tables.

DM v/s ER Models

DM ERUsed to design database for Online Analytical Processing (OLAP)

Used to design database for Online Transaction Processing (OLTP)

Support ad hoc end-user queries Support defined queries

Intuitive & facilitates high-performance retrieval of data

Removes redundancy of data

De-normalized Normalized

Fact Tables• Primary table in the DM

• Each row corresponds to a measurement

• Facts in the fact table are numeric and additive

• Narrow rows with a few columns

• Large number of rows (billions)

• Express many-to-many relationships between dimensions

Dimension Tables• Define business in terms already familiar to users

• Implement the user interface to the DW

• Wide rows with lots of descriptive text

• Small tables (about a million rows)

• Joined to fact table by a foreign key

• Heavily indexed

• E.g. of typical dimensions• time periods, geographic region (markets, cities),

products, customers, salesperson, etc.

Four Step Dimensional Design Process

• Step 1 - Select the business process to model• The first step in converting an ER diagram to a set of

DM diagrams is to separate the ER diagram into its discrete business processes and to model each one separately.

• Step 2 - Choose The Grain of the Business Process

• The grain is the fundamental atomic level of data to be represented in the fact table.

Four Step Dimensional Design Process (cont’d)

• Step 3 - Designate the Fact Tables• The third step is to select those many-to-many

relationships in the ER model containing numeric and additive non-key facts and to designate them as fact tables.

• Step 4 - Choose the dimensions that will apply to each fact table record

• This involves de-normalizing all of the remaining tables into flat tables with single-part keys that connect directly to the fact tables.

Classic Star Schema Model

Snowflake Schema

Fact Constellation Schema

Slowly Changing Dimensions• Type 1: Overwrite the value

Slowly Changing Dimensions (cont’d)• Type 2: Add a Dimension row

• Type 3: Add a Dimension column

Conceptual Modeling

Graph Theory

• Directed, acyclic, weakly connected graph

• Quasi-tree

The Dimensional Fact Model

• Fact Schemes• Facts

• Measures

• Dimensions

• Hierarchies Dimension attributes Non-dimension attributes

The Dimensional Fact Model

Why Formalize?

Why Formalize?

• Give meaning to the model

• Tool support• Transformation Algorithms

• CASE-Tool (Computer Aided Software Engineering)

Fact Scheme SORNAMf ,,,,,

• M is a set of measures

• A is a set of dimension attributes

• N is a set of non-dimension attributes

• R is a set of ordered couples, having the form (ai, aj), indicating the ‘edges’ of the scheme

ji

j

i

aa

NAa

aAa

0


• O is a set of optional relationships

• S is a set of aggregation statements, in the form (mj, di, Ω)

RO

,...,,,,, ORANDMAXCOUNTAVGSUM

fDimd

Mm

i

j


• We call the set Dim(f) a dimension pattern. Each element in Dim(f) is a dimension

RaaAafDim ii ,0


Algorithm

From ER to Conceptual Design

1) Define Facts

2) For each facta) Build attribute tree

b) Prune & Graft

c) Define Dimensions

d) Define Measures

e) Define Hierarchies

Sample Schema

Define Facts• Entity F

• Relationship R between entities E1…En

• Transform R into an entity F

• Frequently updated archives are good candidates for defining facts• E.g. Sale

• Not: Store, City

• Each Fact becomes a root in a fact scheme

Transform Relation

Build Attribute Tree

• Each vertex corresponds to an attribute of the scheme

• Root corresponds to the identifier of F


root=newVertex(identifier(F));

translate(F, root);


translate(E,v) { for each attribute a E | a identifier(E) addChild(v, newVertex({a})); for each entity G connected to E by a relationship R | max(E,R) = 1 { for each attribute b R addChild(v, newVertex({b})); next=newVertex(identifier(G)); addChild(v, next); translate(G, next); }}

Exampletranslate(E=SALE, v=sale)

addChild(v, qty);

addChild(v, unitPrice);

for G=PURCHASE TICKET

addChild(v, ticketNumber);

translate(PURCHASE TICKET, ticketNumber)

for G=PRODUCT

addChild(v, product);

translate(PRODUCT, product);

Attribute Tree

Attribute Tree

• Label the root with the name of the entity F instead of his identifier

• Optional relationships not in algorithm if min(E,R)=0

From ER till Conceptual Design

a) Build attribute tree

b) Prune & Graft

c) Define Dimensions

d) Define Measures

e) Define Hierarchies

Prune & Graft

• Prune or graft to eliminate unnecessary level of detail

• Pruning: Drop a subtree from the quasi-tree

• Grafting: Vertex contains uninteresting information but its descendants must be preserved

Graftgraft(v) {

for each v’ | v’ is father of v

for each v’’ | v’’ is child of v

addChild(v’, v’’);

drop(v);

}

Graft

• 1-to-1 relation is a good candidate

• When an optional vertex is grafted, all his children inherit the optional dash

Prune & Graft

Prune & Graft

Dimensions

• Determines the granularity of fact instances

• Time is a key dimension• Snapshot

• Temporal

Measures

• Numerical attributes of the attribute tree

• Glossary• How measure can be calculated from source

scheme

• e.g. qty sold, no. of customers

Hierarchies

• Tree has already a kind of hierarchy• We can still prune/graft details

• Add new levels for aggregation• E.g. month-quarter-year

• Identify non-dimension attributes• E.g. address

Aggregation

• Primary fact instances• Null assumption

• Zero assumption

• Roll-up

• Sum, Avg, Count, Min, Max, …

Aggregation

• Graphical Notation• Sum

Multi-Aggregation

Multi-Aggregation

• Order matters• {week, product} {month, type}

• Time-Dimension: Min

• Product-Dimension: Sum

Multi-Aggregation

Multi-Aggregation

typemonthtypeweekproductweek MINSUM ,,,



typemonthproductmonthproductweek SUMMIN ,,,



Indexing

Cost Model

• Cost of answering a query is number of rows processed

• Subcubes• Powerset of the dimensions

Cost Model

Indexes

• B-tree indexes to speed up query processing

• E.g. for cube ps, we can construct the following indexes• Ips

• Isp

Example

• Consider Q1:• Using subcube ps: 0,8M rows

• Using subcube psc: 6M rows

• What if we use index Isp on subcube ps?

• 80 rows

sp

s

ps

Indexes

• Ideal situation• All subcubes

• All indexes

Algorithms

• Balance space subcubes – indexes

• Greedy Algorithm• Given a set of queries

• Every step select index/subcube with the highest benefit

?

References• Text books

• Ralph Kimball, The Data Warehouse Toolkit, John Wiley and Sons, 1996

• W.H. Inmon, Building the Data Warehouse, Second Edition, John Wiley and Sons, 1996

• Barry Devlin, Data Warehouse from Architecture to Implementation, Addison Wesley Longman, Inc 1997

• Research Papers/Whitepapers• M. Golfarelli, D. Maio, S. Rizzi, The Dimensional Fact Model: a Conceptual

Model for Data Warehouses, International Journal of Cooperative Information, Vol.7 (issue 2/3), pages 215-247, 1998.

• H. Gupta, V. Harinarayan, A. Rajaraman, J.D. Ullman, Index Selection for OLAP, Proceedings of the Thirteenth international Conference on Data Engineering, April 07 - 11, pages 208-219, 1997.

• S. Luján-Mora J. Trujillo. A comprehensive method for data warehouse design. Proc. DMDW, 2003.

References (cont’d)• Luján-Mora, S., Trujillo, J., and Song, I. Extending the UML for

Multidimensional Modeling. Lecture Notes In Computer Science, Vol. 2460, pages 290-304., 2002.

• Husemann, B., Lechtenborger, J., Vossen, G.: Conceptual Data Warehouse Design.

• In: Proc. of the 2nd. Intl. Workshop on Design and Management of Data Warehouses (DMDW'2000), Stockholm, pages 3-9, 2000.

• Lehner, W., Albrecht, J., and Wedekind, H. 1998. Normal Forms for Multidimensional Databases. In Proceedings of the 10th international Conference on Scientific and Statistical Database Management (July 01 – 03), pages 63-72, 1998.

• Web Articles• http://en.wikipedia.org/wiki/Data_warehouse

• http://en.wikipedia.org/wiki/Online_analytical_processing

• http://en.wikipedia.org/wiki/OLTP

References (cont’d)• http://www.sidadelman.com/data_warehouse_applications.htm

• http://infolab.stanford.edu/infoseminar/Archive/FallY97/slides/ncr

• www.cdd.go.th/it/file/DataWarehousing_and_DataMining.pdf

• http://www.ciobriefings.com/whitepapers/StarSchema.asp

Documents

Data Warehouse Modeling