Você está na página 1de 51

Metadata and Business Intelligence

Feira Internacional de Business Intelligence

So Paulo, Brasil 19 de abril de 2002


1 2002Database Design Solutions, Inc.
40 Morristown Road, Suite 2B Bernardsville, NJ 07924 www.dbdsolutions.com

Lining Up the Data

Connecting the numbers with the brains

2 2002Database Design Solutions, Inc.

40 Morristown Road, Suite 2B Bernardsville, NJ 07924 www.dbdsolutions.com

Important Topics
o Metadata
What it is Its Impact on Business Intelligence (BI)

o The Well-Instituted Metadata Environment o Metadata before BI


How to do it

3 2002Database Design Solutions, Inc.

40 Morristown Road, Suite 2B Bernardsville, NJ 07924 www.dbdsolutions.com

The Ultimate Objective


Data Information

Knowledge

Metadata

Intelligence!

4 2002Database Design Solutions, Inc.

40 Morristown Road, Suite 2B Bernardsville, NJ 07924 www.dbdsolutions.com

So Where is the Knowledge?

Employees Databases

Applications Documents
5 2002Database Design Solutions, Inc.
40 Morristown Road, Suite 2B Bernardsville, NJ 07924 www.dbdsolutions.com

The 5 Questions
o o o o o What data do I have? What does it mean? Where is it? How did it get there? How do I get it? (Go get it for me)

6 2002Database Design Solutions, Inc.

40 Morristown Road, Suite 2B Bernardsville, NJ 07924 www.dbdsolutions.com

More 5 Questions
1. 2. 3. 4. 5. What metadata do I have? What does it mean? Where is it? How did it get there? How do I get it? (Go get it for me)

7 2002Database Design Solutions, Inc.

40 Morristown Road, Suite 2B Bernardsville, NJ 07924 www.dbdsolutions.com

Getting to Business Intelligence


Data Information

Knowledge

Metadata

Intelligence!

8 2002Database Design Solutions, Inc.

40 Morristown Road, Suite 2B Bernardsville, NJ 07924 www.dbdsolutions.com

Business Intelligence
o What do we have? o How much? o What have we done?
When? How often?
What resulted? Was it worthwhile?

9 2002Database Design Solutions, Inc.

40 Morristown Road, Suite 2B Bernardsville, NJ 07924 www.dbdsolutions.com

Can WE Derive Decisions From BI?


Data
Information

Metadata

10 2002Database Design Solutions, Inc.

40 Morristown Road, Suite 2B Bernardsville, NJ 07924 www.dbdsolutions.com

Not Without Metadata!


o The Way In o The Way Around o The Way Out

11 2002Database Design Solutions, Inc.

40 Morristown Road, Suite 2B Bernardsville, NJ 07924 www.dbdsolutions.com

Business Intelligence?
o Without information, which stems from data and process, businesses would be far from intelligent

12 2002Database Design Solutions, Inc.

40 Morristown Road, Suite 2B Bernardsville, NJ 07924 www.dbdsolutions.com

What is Metadata?

2002 Database Design Solutions, Inc.

40 Morristown Road, Suite 2B Bernardsville, NJ 07924 www.dbdsolutions.com

Information vs. Knowledge


o Today, nothing becomes information unless some processing or context, usually of a particular specialty, is applied to it.
o Information evolves.
Attendance List Student Name

Art
Report Card Course Name

14 2002Database Design Solutions, Inc.

40 Morristown Road, Suite 2B Bernardsville, NJ 07924 www.dbdsolutions.com

Metadatas Role
J5043

Instance Data

Characters without Context

Metadata
Part Inventory Guide

Context

J5043
Part Number for Windshield Wiper

Information

Information

15 2002Database Design Solutions, Inc.

40 Morristown Road, Suite 2B Bernardsville, NJ 07924 www.dbdsolutions.com

Putting Metadata Into Perspective


o We wouldnt be able to survive without it Unless we already experienced or interpreted this data

16 2002Database Design Solutions, Inc.

40 Morristown Road, Suite 2B Bernardsville, NJ 07924 www.dbdsolutions.com

The Map Analogy


o Instance Data: The Map o Metadata: The Legend

17 2002Database Design Solutions, Inc.

40 Morristown Road, Suite 2B Bernardsville, NJ 07924 www.dbdsolutions.com

The Supermarket Analogy


o Aisle markers
Incorrect aisle markers Missing aisle markers

18 2002Database Design Solutions, Inc.

40 Morristown Road, Suite 2B Bernardsville, NJ 07924 www.dbdsolutions.com

The Product Assembly Analogy


o Ever try to put something together without instructions?
Or with incorrect instructions? Or with instructions that make no sense? How about with pieces that do not have part numbers?

19 2002Database Design Solutions, Inc.

40 Morristown Road, Suite 2B Bernardsville, NJ 07924 www.dbdsolutions.com

Metadata Confirmed
o It identifies what we want
What data do we have?

o It clarifies what we want


What does it mean?

o It locates what we want


Where is it?

o It validates what we want


Where did it come from?

o It gets what we want


How do I get it? Go get it for me!
20 2002Database Design Solutions, Inc.
40 Morristown Road, Suite 2B Bernardsville, NJ 07924 www.dbdsolutions.com

Comparing Data to Metadata


o We use metadata (card catalogs, library search sites) to list the data (books, periodicals, videos, etc.) o We use metadata (titles, indices, tables of contents) to understand the data (text, images,etc.) o We use metadata (ISBN numbers, shelves) to locate the data (books, periodicals, videos, etc.) o We use metadata (author, publication date, publisher) to source the data (books, periodicals, videos, etc.) o We use metadata (ISBN numbers, shelves) to access the data (books, periodicals, videos, etc.)

The Library Analogy

21 2002Database Design Solutions, Inc.

40 Morristown Road, Suite 2B Bernardsville, NJ 07924 www.dbdsolutions.com

Data/Metadata/Information
Data: $46.30 o o o o o What data do I have? What does it mean? Where is it? How did it get there? How do I get it? o Product_price o The gross charge for the firms offerings, before any discounts o Sales Database Product Table o Update_Product Program o Product_price Query
22 2002Database Design Solutions, Inc.
40 Morristown Road, Suite 2B Bernardsville, NJ 07924 www.dbdsolutions.com

Metadatas Impact on BI
o Consider finding all of the data that you need to make a decision o Consider interpreting found data o Consider basing new decisions on past trends

23 2002Database Design Solutions, Inc.

40 Morristown Road, Suite 2B Bernardsville, NJ 07924 www.dbdsolutions.com

Where Should the Metadata Come From?


o Is it already around? o Is it worth anything? o Should you create it all over again? o How do you make the right decision?

24 2002Database Design Solutions, Inc.

40 Morristown Road, Suite 2B Bernardsville, NJ 07924 www.dbdsolutions.com

The Well Instituted Metadata Environment


What it is and how to get there

2002 Database Design Solutions, Inc.

40 Morristown Road, Suite 2B Bernardsville, NJ 07924 www.dbdsolutions.com

A Metadata Web
BI Tool Middle ware

Decision Support 'Front End Report Name Table Alias Report Column Table Name Report Element Name Network Address Report Element Calculation User Id Element Format Database Name Userid Virtual Table Name Report Security Server Name Data Element Name ODBC Name Data Element Location Data Warehouse Name

Database Name Table Name Column Name Index Name Query Name Userid
Data

ETL Tool

Database Name Table Name Column Name Data Index Name Query Name Userid
Data

Source Data Store Name Source Table Name Source Column Name Translation Process Name

Database Name Table Name Column Name Index Name Query Name Userid ..

26 2002Database Design Solutions, Inc.

40 Morristown Road, Suite 2B Bernardsville, NJ 07924 www.dbdsolutions.com

How We Should Handle Our Metadata (Today?)


Unique Development Metamodel Constructs Unique Reporting Tool Metamodel Constructs

Common Metamodel Constructs

Metadata Database Reporting Tool

The Integrated Metamodel


Metadata Database DBMS Catalog Development Tool

Metadata Database

27 2002Database Design Solutions, Inc.

40 Morristown Road, Suite 2B Bernardsville, NJ 07924 www.dbdsolutions.com

Well Instituted Means


o The metadata helps instead of hinders o Disparate metadata is logically organized o Multiple metadata stores are possible, related, and accessible.
These metadata stores can be part of different vendor products Architectural access points are common

o Individuals do not need to access an entire architecture to reach metadata that is specific to their world.
28 2002Database Design Solutions, Inc.
40 Morristown Road, Suite 2B Bernardsville, NJ 07924 www.dbdsolutions.com

Worthwhile Metadata?
o Must have 5 stars
Accuracy Consistency Modularity Represented Perspectives Ease of Identification

29 2002Database Design Solutions, Inc.

40 Morristown Road, Suite 2B Bernardsville, NJ 07924 www.dbdsolutions.com

Becoming Well Instituted


o Scope your Business Intelligence World
Information Sources Involved Tools/Products Metadata Requirements

30 2002Database Design Solutions, Inc.

40 Morristown Road, Suite 2B Bernardsville, NJ 07924 www.dbdsolutions.com

The Need For Requirements


o Requirements = Planning. o Needed for Metadata Solutions as well as for Data (Application) Solutions. o Dont build it before you know what it is being built for! o Dont compromise requirements to match a particular vendor solution

31 2002Database Design Solutions, Inc.

40 Morristown Road, Suite 2B Bernardsville, NJ 07924 www.dbdsolutions.com

Requirements Approach
Data Requirements Metadata Requirements o Requirements come o Requirements come from users (people) from users (people, tools, other metadata o Some data could be solutions) new o Metadata already exists

Therefore, identifying Metadata Requirements requires a different approach


32 2002Database Design Solutions, Inc.
40 Morristown Road, Suite 2B Bernardsville, NJ 07924 www.dbdsolutions.com

The Metadata Requirements Analysis Process


o Needed for both small standalone Metadata solutions and large Enterprise wide solutions.

33 2002Database Design Solutions, Inc.

40 Morristown Road, Suite 2B Bernardsville, NJ 07924 www.dbdsolutions.com

Requirements Process Steps


1. Identify Metadata Beneficiaries (i.e., users, e.g, developers). 2. List metadata requirements (e.g., data element definition) by beneficiary. 3. Determine the source of each metadata requirement (e.g, data dictionary).
For multiple sources, identify the metadata of record.

34 2002Database Design Solutions, Inc.

40 Morristown Road, Suite 2B Bernardsville, NJ 07924 www.dbdsolutions.com

Requirements Process Steps contd


4. Plan preliminary architecture and metadata solution scope. 5. Re-categorize metadata. Determine the usage category (i.e., specific, unique, or common) of each metadata requirement. 6. Build metamodels; models of the metadata.

35 2002Database Design Solutions, Inc.

40 Morristown Road, Suite 2B Bernardsville, NJ 07924 www.dbdsolutions.com

Requirements Process Steps contd


7. Consider access (i.e., CRUD = Create, Read, Update, Delete) and display requirements for metadata. 8. Anticipate metadata processing requirements (usually specific to software) 9. Consider non-metadata requirements (technical support, security, organization, etc.)
36 2002Database Design Solutions, Inc.
40 Morristown Road, Suite 2B Bernardsville, NJ 07924 www.dbdsolutions.com

A Multi-Tool Architecture
Metadata Database BI Tool Metadata Database

Metadata Database

DBMS Catalog

Metadata Database

Development Tool

37 2002Database Design Solutions, Inc.

40 Morristown Road, Suite 2B Bernardsville, NJ 07924 www.dbdsolutions.com

A Metadata Solution Architecture


Report

DBMS API Search and Retrieve Internet Search Engines


ProfileControlled Format Metadata Database BI Tool Column Source Data Element

Metadata Database

Transformation Source Program Source Data Element

Initial Batch Load

DBMS APIs
Database DBMS Catalog Table Metadata Database
38 2002Database Design Solutions, Inc.

XML Tagged Exchange


Development Tool

Metadata Database

Column

40 Morristown Road, Suite 2B Bernardsville, NJ 07924 www.dbdsolutions.com

Sharing Metadata?
o For Business Intelligence to benefit from a well-instituted metadata solution:
Each tool component has the ability to access metadata from outside of its own database Non-tool users can access metadata within a particular tools database

39 2002Database Design Solutions, Inc.

40 Morristown Road, Suite 2B Bernardsville, NJ 07924 www.dbdsolutions.com

If Its Good, Everyone Wants It


o Just make sure everyone can get it!

40 2002Database Design Solutions, Inc.

40 Morristown Road, Suite 2B Bernardsville, NJ 07924 www.dbdsolutions.com

Metadata Before BI?


What comes first?

2002 Database Design Solutions, Inc.

40 Morristown Road, Suite 2B Bernardsville, NJ 07924 www.dbdsolutions.com

If BI Comes First
o We put something together before reading the directions o We go food shopping without a list o We buy clothes without trying them on o We cook an egg and then try to separate it

42 2002Database Design Solutions, Inc.

40 Morristown Road, Suite 2B Bernardsville, NJ 07924 www.dbdsolutions.com

If Metadata Comes First


o o o o We travel with directions We cook with recipes We shop with lists We try before we buy

Everything is performed: More efficiently With few errors In substantially less time!

43 2002Database Design Solutions, Inc.

40 Morristown Road, Suite 2B Bernardsville, NJ 07924 www.dbdsolutions.com

Supporting BI With Metadata

Getting Started

44 2002Database Design Solutions, Inc.

40 Morristown Road, Suite 2B Bernardsville, NJ 07924 www.dbdsolutions.com

Sealing Your Requirements


1. Determine the data to be used for BI
Current Data Stores Original Sources Transitions Along the Way Does it answer the 5 Questions What data do I have (available for BI)? What does it mean? Where is it (within my BI world)? How did it get there (sourcing)? How do I get it (BI tool query, etc.)

2. Evaluate its associated metadata (if you have it!)

45 2002Database Design Solutions, Inc.

40 Morristown Road, Suite 2B Bernardsville, NJ 07924 www.dbdsolutions.com

Sealing Your Requirements - contd


3) Organize Metadata By Defined Data and by Beneficiary Interest
o Metadata which answers 5 questions about CRM data
o Subsets of above metadata based upon who will need it and for what purpose

4) Develop Metamodel as needed 5) Populate Metamodel

46 2002Database Design Solutions, Inc.

40 Morristown Road, Suite 2B Bernardsville, NJ 07924 www.dbdsolutions.com

Sealing Your Requirements - contd


4) Assure Metadata Accessibility within BI architecture

47 2002Database Design Solutions, Inc.

40 Morristown Road, Suite 2B Bernardsville, NJ 07924 www.dbdsolutions.com

Resulting Metadata
o Will address the data within the initial BI scope o Will have an associated metamodel that is flexible enough to cover additional data both within and outside of the BI environment o Will be as accessible as the BI data itself

48 2002Database Design Solutions, Inc.

40 Morristown Road, Suite 2B Bernardsville, NJ 07924 www.dbdsolutions.com

Conclusion
Metadata and BI should live together

2002 Database Design Solutions, Inc.

40 Morristown Road, Suite 2B Bernardsville, NJ 07924 www.dbdsolutions.com

BI Without Metadata.
o Should be against your religion o Will make your life miserable (eventually) o Does not justify the time and expense involved in setting up a BI environment

50 2002Database Design Solutions, Inc.

40 Morristown Road, Suite 2B Bernardsville, NJ 07924 www.dbdsolutions.com

Questions?
o Recent Book:
Metadata Solutions: Using Metamodels, Repositories, XML, and Enterprise Portals to Achieve Information on Demand (2001, Addison Wesley)

o Visit our web site:


www.dbdsolutions.com Publications Discussion Group

o Send me email
atannenbaum@dbdsolutions.com

51 2002Database Design Solutions, Inc.

40 Morristown Road, Suite 2B Bernardsville, NJ 07924 www.dbdsolutions.com

Você também pode gostar