Teradata Free Tutorial
- tekslate
- Jan 3, 2015
- 3 min read
Teradata Corporation is a global leader in providing data warehousing technologies. Almost all of the largest data warehouses on earth use Teradata RDBMS. Teradata was formerly a division of NCR Corporation with their headquarters in Dayton. They spin-off from NCR on October 1, 2007 and now a truly technology oriented company.
Teradata is a complete relational database management system with its massively parallel processing mechanism running on distributed computing architecture which is scalable in all dimensions of a database system. The beauty of Teradata is that it supports both Symmetric Multi-Processing (SMP) and Massively Parallel Processing (MPP) systems where the distributed functions communicate by means of a fast interconnect structure known as BYNET.
Teradata Components
Passing Engine
It takes SQL Request and delivery SQL response
Session Control
Session is nothing but logical connection between user and Application. Passing Engine checks the authentication and authorization, if it’s valid, it does log in other wise log off
Parser
It checks syntactical Error
It checks symmetrical Error
It checks Existence of object
Query Rewrite Mechanism
It writes the Query such away optimizer can understand easily
Example
IF The Query is Q7,it converts into Q2
It is important component in any database including Teradata
It provides execution plan for SQL Statement, so that it is going to be executed by database
Generally it usages the below information.
System configuration
Available parallel unit
Data Demographics, etc
Access plan
Join plan
Generator
It generated steps for plan provided
Dispatched
It performs and manager request and response, flow control(Taking Request, Response keeping in Queue in delivery)
Note
Passing Engine Handle max 120 sessions at a time
It acts like” Message communication layer” between various components.
SMP BYNET [PE-AMP]
POINT - POINT-One message from PE To one AMP
MULIT CAST-One message from PE to Many AMPs
BROAD CAST- One message from PE to All AMPs
MPP BYNET[NODE-NODE]:
POINT- POINT-One message from One NODE to other NODE
MULIT CAST-One message from One NODE to Many NODE
BROAD CAST- One message from One NODE to All NODE
AMP[Access Module Processor]
Each AMP is responsible for managing a portion of MAIN DISK SPACE (VIRTUAL DISK)
This space not sharable by any other AMPs so we call architecture as shared nothing architecture.
Each AMP Operator independently resources
Each AMP contain database management sub system. If perform the below operations.
Performing DDL
Performing DML
Performing Joins
Performing Applying and Releasing locks
Performing Aggregate operation
Performing OLAP operation
Performing SORTING Operation etc.,
Note
Max so task AMP TO perform at a time
Multiple AMP runs independently so high parallelism implement
VSS(Virtual Storage System):
It is available from teradata 13 onwards, basically design to manage a multi tier ware house
Teradata VSS Pools all of the cylinders with in CLIQUE’s DISK SPACE and allocates cylinders from storage pool.
It migrates frequently used data (Hot data) to FASTER DEVICES and less frequently used data (cold data)to SLOWER DEVICES.
Before Teradata VSS
1. a) AMPS know the physical location of cylinder which are address by DRIVE# (OR) CYLINDER#
2. b) Adjusting system AMPS integral number of drives per AMP
3. c) Adding storage requires an additional drive per AMP.
After TDV SS
1. a) AMPs don’t know the physical location of cylinder and it can be change, because of the cylinder in CLIQUE’s On effective in pool, that are managed by teradata virtual storage, virtual process.
2. b) Added drives are shared by all AMP’s
3. c) You can add a number of drives, This new Drive may have difference capacity are performance than those drives, which are already present in the system.
To Learn More Click On Below Link:
Comments