2. Introduction
• The Past and The Problem
• What is a Data Warehouse?
• Components of a Data Warehouse
• OLAP, Metadata, Data Mining
• Getting the Data in
• Benefits vs. Costs
• Conclusion & Questions
3. The Past and The Problem
• Only had scattered transactional systems in the
organization – data spread among different
systems
• Transactional systems were not designed for
decision support analysis
• Data constantly changes on transactional systems
• Lack of historical data
• Often resources were taxed with both needs on the
same systems
4. The Past and The Problem
• Operational databases are designed to keep
transactions from daily operations. It is
optimized to efficiently update or create
individual records
• A database for analysis on the other hand
needs to be geared toward flexible requests
or queries (Ad hoc, statistical analysis)
5. What is a Data Warehouse?
Data warehousing is an architectural model
designed to gather data from various
sources into a single unified data model for
analysis purposes.
6. What Is a Data Warehouse?
Term was introduced in 1990 by William
Immon
A managed database in which the data is:
• Subject Oriented
• Integrated
• Time Variant
• Non Volatile
7. Subject Oriented
• Organized around major subject areas in the
enterprise (Sales, Inventory, Financial, etc.)
• Only includes data which is used in the
decision making processes
• Elements used for transactional processing are
removed
8. Integrated
• Data from different sources are brought
together and consolidated
• The data is cleaned and made consistent
Example – Bank Systems using Different Codes
Loan Department – COMM
Transactional System - C
9. Time Variant
• Data in a Data Warehouse contains both
current and historical information
• Operational Systems contain only current
data
Systems typically retain data:
Operational Systems – 60 to 90 Days
Data Warehouse – 5 to 10 Years
10. Non Volatile
• Operational systems have continually
changing data
• Data Warehouses continually absorb current
data and integrates it with its existing data
(Aggregate or Summary tables)
Example of volatile data would be an account balance at a bank
11. What Is a Data Warehouse?
• Not a product, it is a process
• Combination of hardware and software
• Concept of a Data Warehouse is not new,
but the technology that allows it is
12. What Is a Data Warehouse?
Can often be set up as one VLDB (Very Large
Database) or a collection of subject areas
called Data Marts.
There are now tools which “unify” these Data
Marts and make it appear as a single
database.
13. What Is a Data Warehouse?
Transformation of Data to Information
Transaction Processing
Cleansing / & Normalization
Relational Warehouse
SQL reporting
Exploration / Analysis
Data
Information
14. Components of a Data Warehouse
Four General Components:
• Hardware
• DBMS - Database Management System
• Front End Access Tools
• Other Tools
In all components scalability is vital
Scalability is the ability to grow as your data and processing needs
increase
15. Components of a Data Warehouse -
Hardware
• Power - # of Processors, Memory, I/O Bandwidth,
and Speed of the Bus
• Availability – Redundant equipment
• Disk Storage - Speed and enough storage for the
loaded data set
• Backup Solution - Automated and be able to allow
for incremental backups and archiving older data
16. Components of a Data Warehouse -
DBMS
• Physical storage capacity of the DBMS
• Loading, indexing, and processing speed
• Availability
• Handle your data needs
• Operational integrity, reliability, and
manageability
17. Components of a Data Warehouse -
Front End & Other Tools
• Query Tools (SQL & GUI based)
• Report Writers
• Metadata Repositories
• OLAP (Online Analytical Processing)
• Data Mining Products
18. Components of a Data Warehouse –
Metadata Repositories
Metadata is Data about Data. Users and Developers
often need a way to find information on the data
they use. Information can include:
• Source System(s) of the Data, contact information
• Related tables or subject areas
• Programs or Processes which use the data
• Population rules (Update or Insert and how often)
• Status of the Data Warehouse’s processing and
condition
19. Components of a Data Warehouse –
OLAP Tools
OLAP - Online Analytical Processing. It works by
aggregating detail data and looks at it by dimensions
• Gives the ability to “Drill Down” in to the detail data
• Decision Support Analysis Tool
• Multidimensional DB focusing on retrieval of
precalculated data
• Ends the “big reports” with large amounts of detailed data
• These tools are often graphical and can run on a “thin
client” such as a web browser
20. Components of a Data Warehouse –
Data Mining
• Answers the questions you didn’t know to
ask
• Analyzes great amounts of data (usually
contained in a Data Warehouse) and looks
for trends in the data
• Technology now allows us to do this better
than in the past
21. Components of a Data Warehouse –
Data Mining
• Most famous example is the Huggies -
Heineken case
• Used in Retail sector to analyze buying
habits
• Used in financial areas to detect fraud
• Used in the stock market to find trends
• Used in scientific research
• Used in national security
22. Getting the Data In
• Data will come from multiple databases and
files within the organization
• Also can come from outside sources
• Examples:
• Weather Reports
• Demographic information by Zip Code
23. Getting the Data In
Three Steps :
1. Extraction Phase
2. Transformation Phase
3. Loading Phase
24. Getting the Data In
Extraction Phase:
• Source systems export data via files or
populates directly when the databases can
“talk” to each other
• Transfers them to the Data Warehouse
server and puts it into some sort of staging
area
25. Getting the Data In
Transformation Phase:
• Takes data and turns it into a form that is suitable
for insertion into the warehouse
• Combines related data
• Removes redundancies
• Common Codes (Commercial Customer)
• Spelling Mistakes (Lozenges)
• Consistency (PA,Pa,Penna,Pennsylvania)
• Formatting (Addresses)
26. Getting the Data In
Loading Phase:
• Places the cleaned data into the DBMS in
its final, useable form
• Compare data from source systems and the
Data Warehouse
• Document the load information for the users
28. Benefits
• Creates a single point for all data
• System is optimized and designed
specifically for analysis
• Access data without impacting the
operational systems
• Users can access the data directly without
the direct help from IT dept
29. Costs
• Cost of implementation & maintenance
(hardware, software, and staffing)
• Lack of compatibility between components
• Data from many sources are hard to
combine, data integrity issues
• Bad designs and practices can lead to costly
failures
30. Conclusion
• What is a Data Warehouse?
• Components of a Data Warehouse
• How the Data Gets In
• OLAP, Metadata, and Data Mining
• Benefits vs. Costs
Welcome
Honor to give back to PSU
Resume
Announce Informal Rules
Review Slide
Last Bullet - Resources being taxed is the lead in to the next slide
Operational Systems – Many Small Writes,
Data Warehouse – Heavy “Off” hour loads – Also can have a “trickle feed”
Go over part that is in pink. - The only thing for them to remember
Word “warehouse” – Reminds me of the end Indiana Jones when they had the Ark and put it in the large government warehouse never to be seen again. It implies storage.
Data Warehouse is not only for business. Military, Intelligence, and Scientific Needs:
9/11 & Human Genome Project
Lead in to next 4 slides
Examples of what is removed is control fields and timestamps needed for transactional processing.
Additionally Elements could be too small for analysis purposes – Retail environment recording the opening of a cash register drawer.
Example with Bank is Commercial Customers
Goal is for the data from the different sources to work seamlessly together.
Operational systems usually need to purge some of this data in order for lookups to be efficient.
Additional Hardware Costs as well
Depending on the time of day the balance can change. Difficult to make decisions.
DW – The data is usually checked and finalized. Once the data is finalized it should not change.
Corrections of errors do happen after the fact but the should be communicated to the users.
Despite marketing claims, can’t just buy a Data Warehouse
Collection of tools and processes
Improvement in Processing and Lower costs for this technology
Data Marts can be divided up by subject areas or geographically
Data Marts tend to be small departmental specific
Geographic division is an example of Distributed System.
Users don’t know where the data is coming from. In many cases, they don’t need to. However there should be documentation that the users can get to.
Give example of 20,000 is data?
What does it mean? Sales In The West of BBQ Grills in 2000?
Now if I say Wilt Chamberlain – Something completely different comes to mind.
Data Warehousing is a step taking low level transactional data and turning it into useful information.
Growth is important, to switch to another solution can be very costly.
Data – More years of data from more subject areas
Processing – Load times, Number of users, Query or request times.
Power – Requires powerful machines such as a main frame or clusters of servers like UNIX that can work together
Storage – Not only disk, there is tape and optical
Data Needs – No longer values and text. Searchable Pictures, fingerprint, video, audio, documents. Someday DNA.
Operational integrity – Enforce its rules – Security and “Atomic Transactions”
Reliability - Can it recover from failure quickly and easily
Manageability – Ability to do the day to day tasks with little or no effort.
Last 3 we will go into more detail
Query Tools – Not uncommon to have different tools for different types of users – Power Users
Report Writers – Prepare and distributed to many different mediums (email,web,wireless,voice).
Can be electronic and centrally stored also.
Value of metadata is sometimes over looked – Cost of Man hours.
Column names alone are not enough. Field named “Date”
Organizations without this sometime spend a lot of time researching this.
Changes in Data – Impact is Known
Example: White Board:
Dimensions represent the core components of a business plan and often relate to departmental functions. Typical standard dimensions are Time, Accounts, Product Line, Market, and Division. Dimensions are static in most databases; database dimensions rarely change over the life of the application.
Dimensions
Organization – East,West
Time – 2000, 2001
Product – Gas Grills, Charcoal Grills
Technology Will Only Improve Data Mining
Better connectivity is eliminating the need for raw files
Let users know what data got updated
Very High Maintenance:
Staffing & Roles
DBAs, Data Architects, Hardware/OS Specialists, Data Auditors, Support for Front End Tools, Documentation, Security
