Intr Teradata

7/28/2019 Intr Teradata

1/17

Introduction to

Teradata

Kanakadurga Rajanala


2/17

Overview

What is a Independent Data Mart?

What is Data Warehousing?

What is Teradata?

Teradata Architecture

Data Storage and Retrieval

Fault Tolerance Q & A


3/17

Independent Data Marts

Engg Dept

Decision

SupportInformation

Arts Dept

Decision Support

Information

Science Dept

Decision

SupportInformation

How to get university wide reports?

How to prevent data redundancy?

How to maintain consistent data across departments?


4/17

What is a Data Warehouse?

A data warehouse is a central, enterprise wide

database that contains information obtained

from operational systems.

Good Data warehouse

Provides required storage capability

Performance

Scalability


5/17

Data Warehouse

Engg Dep

Etc

Arts Music

Science

Data

Warehouse

University

Database

Student/faculty info

Copied,

organized

detailed

Data Mining


6/17

What is Teradata?

Teradata is an RDBMS designed for enterprise

data warehousing.

Massively Parallel Processing system(MPP) Parallelism throughout Platform

Share Nothing architecture

Linear Scalability


7/17

Teradata MPP Architecture Physical layout of 4 node system

Each Node is an SMP System. The maximum nodes can be up to 512.

MPP System


8/17

BYNET Interconnect BYNET high-speed interconnect facilitates system

communication

All nodes connected via BYNET

Hardware network

Driver Software runs on each node


9/17

Node Architecture

Each Teradata Node is made up of hardware and software Each node has CPUs, system disk, memory and adapters

Each node runs copy of OS and database SW


10/17

Node Architecture

PE vproc

AMP

vproc

Vdisk

AMP

vproc

Vdisk

AMP

vproc

Vdisk

AMP

vproc

Vdisk

AMP

vproc

Vdisk

AMP

vproc

Vdisk

AMP

vproc

Vdisk

AMP

vproc

Vdisk

PE vproc

Operating System

PDE

V2 Virtual Processors (Vprocs)

Each node has virtual processors (process)

PDE layer facilitates communication between OS andVprocs.

Syntaxer

Optimizer

Step Gen

Dispatch

Execution

EnginesRun in

Parallel

No one

elseshares


11/17

Data Storage

Bucket #

Row Hash

Parsing Engine

Pick Up Primary Index value

Message Passing Layer (BYNET)

AMP AMP AMP AMP

Hash Map

Hashing Algorithm

25

Node-1

Node-2

26

Defined by

user. Used for

data

distribution.

Array

associates

hash buckets

to AMPs


12/17

Client

Data Retrieval

Message Passing Layer

AMP AMP AMP AMP

25

Select * from CalStateLADB.Students;

2910

50

75

Parsing Engine

Dispatcher

29 2510

75

50

RET Step

RET StepRET StepRET StepRET Step

Select * from CalStateLADB.Students;


13/17

Teradata Scales Linearly

Scaling achieved via shared nothing

architecture and unconditional parallelism

Power is in linear scalabilitywhere slope = 1

More nodes

More work

More users

More data

Node

Work

Users

Data

Node1 Node2 Node3

BYNET

Node4


14/17

Fault Tolerance

BYNET

AmpAmp Amp Amp AmpAmpAmpAmp

CLIQUE-1 CLIQUE-2

Disk Array Disk Array Disk Array Disk Array


15/17

Summary

Teradata is designed and used for

enterprise data warehousing.

MPP with Share Nothing Architecture.

Parallelism throughout the platform.

Runs on windows and Linux.


16/17

References

Teradata Basics and Architecture user

manual (www.info.ncr.com)

Teradata Software Demo CD can beordered (free copy) from

www.teradata.com
http://www.info.ncr.com/http://www.teradata.com/http://www.teradata.com/http://www.info.ncr.com/


17/17

Questions?
http://images.google.com/imgres?imgurl=http://webs.wichita.edu/depttools/depttoolsmemberfiles/accomp/question_mark%2520(WinCE).jpg&imgrefurl=http://www.mywackospace.com/blog/2007/06/page/2/&h=320&w=304&sz=9&hl=en&start=87&um=1&tbnid=hk4biq78WeRp3M:&tbnh=118&tbnw=112&prev=/images%3Fq%3Dquestion%2Bmark%2Bicon%26start%3D80%26ndsp%3D20%26svnum%3D10%26um%3D1%26hl%3Den%26rlz%3D1T4GFRC_enUS205IN206%26sa%3DN

Intr Teradata

Documents

Transcript of Intr Teradata