Which compression format is supported by NOS?

Teradata Native Object Store (NOS) supports reading compressed data directly from external object stores using the READNOS table operator. The COMPRESSION parameter within this function explicitly specifies the supported decompression codecs. According to official Teradata Vantage documentation, the valid values for this parameter include 'ZLIB' and 'SNAPPY'. These formats can be used to decompress data from file types like CSV and JSON. While NOS can read complex file formats like Parquet and ORC which have their own internal compression schemes (often using Snappy or ZLIB), the question specifically asks for supported compression formats that can be directly specified.

A. ORC: This is a columnar file format, not a compression format. NOS supports reading ORC files, but ORC itself is a data storage structure.

D. BZIP: This compression format is not listed as a supported value for the COMPRESSION parameter in the official READNOS function documentation.

1. Teradata Vantage™ SQL Functions, Operators, Expressions, and Predicates, Release 17.20. (September 2022). Document ID: B035-1145-172K.

Section: Chapter 2: Table Operators > READNOS > Syntax > COMPRESSION Clause.

Content: This section explicitly lists the supported compression types for the COMPRESSION parameter as 'GZIP', 'SNAPPY', and 'ZLIB'. It does not list 'BZIP'.

2. Teradata Vantage™ - Native Object Store Getting Started Guide, Release 17.20. (September 2022). Document ID: B035-1214-172K.

Section: Chapter 2: Reading Data from External Object Storage > Reading CSV Data > Examples of Reading Compressed CSV Data.

Content: The examples demonstrate the use of COMPRESSION('GZIP') and COMPRESSION('SNAPPY'). This guide also clarifies that for Parquet and ORC, compression is handled automatically based on the file's metadata, distinguishing the file format (like ORC) from the compression codec (like Snappy).