17.00 - Terminology - Teradata Database

Teradata Vantageā„¢ - Native Object Store Getting Started Guide

Advanced SQL Engine
Teradata Database
September 2020
Programming Reference
Files and Objects
Files and objects can be used interchangeably to describe the components of an object store. Each file or object contains records in which the data itself is held.
Foreign Table
Foreign tables carry information about the location of the external object store, as well as other definitional information, and are the main vehicle within Advanced SQL Engine for reading external data. A foreign table may support access to the entire object store, or a subset of an object store.
Key and Path
A key in object store data is the unique identifier for an object and may contain several logical levels.

Path names are identifiers or pointers to an object; a path name is the entire key.

A path prefix is a subset of the path. For example, if the path of an object is /td-usgs/bucket/a/b/c/d/object1, then a path prefix can be any of the following, amongst others:
  • a/b/c
  • a/b
  • a/b/c/d/object1
Object and Objects stores
Objects are the discrete units that compose an object store. Objects can be organized with shared names called prefixes.

Every object has a unique key or path. However, many objects might be identified by or share a common path prefix. For example, /a/b/c/d can contain hundreds of objects.

An object store is a collection of related objects, with all participating objects located in the same bucket or container and organized in a hierarchy.
Records and Rows
Row is a relational concept that refers to part of a table.
Record is a logical grouping of values within an object.
For CSV and JSON formatted data, the database will either convert each record into a row, or will convert a record with an array at the top level into a set of rows.
For Parquet formatted data, fields are grouped together (similar to columnar). Then individual field values are converted to a Teradata column value.