Teradata Architecture

Teradata acts as a single data store, with multiple client applications making inquiries against it concurrently.

Instead of replicating a database for different purpose, with Teradata you store the data once and use it for all clients.

It provides the same connectivity for an entry-level system as it does for a massive enterprise data warehouse.

A Teradata system contains one or more nodes

A node is a term for a processing unit under the control of a single operating system.

The node is where the processing occurs for the Teradata Database there are 2 types of Teradata systems

Symmetric multiprocessing (SMP): An SMP Teradata system has a single node that contains multiple CPU’s sharing a memory pool.

Massively parallel processing (MPP): Multiple SMP nodes working together comprise a larger, MPP implementation of Teradata.

The nodes are connected using BYNET, which allows multiple virtual processors on multiple nodes to communicate with each other.

3 main components

Parsing Engine

BYNET

Access Module Processor