Contents
Chapter 1 – Teradata for Executives
Who is the Most Important Person in a Data Warehouse?
The Basics of a Single Computer
Each AMP is Responsible for a Portion of the Data
Each AMP holds a Portion of Every Table
Understand the Extremes of Teradata
Teradata has TASM – A Sophisticated Traffic System
There are Four Major Types of Users
Chapter 2 – How Teradata Utilizes Parallel Processing
Teradata Stores Data in Tables
Each AMP is Assigned Specific Rows
Each AMP Organizes the Rows inside a Data Block
AMPs Always Transfer Their Data Blocks to Memory
As Tables Get Bigger the AMP uses Multiple Data Blocks
AMPs Process A Table One Block at a Time
The Slowest Processing is a Full Table Scan
Teradata Systems Can Grow Forever
Teradata has Five Designs to Prevent the Full Table Scan
Chapter 3 – How Teradata is Designed
Teradata is Designed Around the Primary Index
Why the Primary Index Important to Teradata? Distribution
Why Else the Primary Index Important to Teradata? Retrieval
Use the Primary Index and Only One Block is Transferred
Why Else the Primary Index Important to Teradata? Sorting
AMPs Know Which Block Holds the Requested Data
As An Executive Know That Teradata Delivers in One Second
Teradata Has a Different Design for Tables Called Partitioning
An All AMPs Retrieve By Way of a Single Partition
Partitioning Uses All AMPs, but a Portion of the Data Moves
What does a Columnar Table look like?
Teradata Has a Different Design for Tables Called Columnar
A Columnar Table is Like Separating Columns in Excel
A Columnar Table is Best for Queries with Few Columns
A Comparison of Data for Normal Vs. Columnar
Which Move From Disk to Memory Would You Choose?
Teradata has Two More Tricks to Move Less Blocks
Teradata Tracks the Most Used Data and Keeps it In-Memory
Teradata Has Secondary Indexes
Chapter 4 – Good Executive Advice
To Understand the Future Learn From the Past
Let the Business Users Decide on How Long to Keep Data
If You Want to Guarantee Success Involve the Business Users
Invest in IT and the Leadership will Pay Big Dividends
Support Large Queries, but Monitor them Closely
Experiment and Improve Loading Data Strategies
Compress Your Data with Multi-Value Compression
Separate your Production System From Your Test System
Have A Multi-Vendor Data Warehouse
Give your Enterprise the Tools they Need
Educate the Business on the Business by Sharing the Model
Load Your Models and have the SQL Built Automatically
Allow the Users to See the Physical Indexes in Their Views
Chapter 5 – Conclusion: The Five Brilliant Pieces of Teradata
Five Brilliant Pieces of Teradata (1 of 5) is MPP
Five Brilliant Pieces (2 of 5) are Tactical Queries
Five Brilliant Pieces (3 of 5) Is a Traffic System