15.00 - Default Definitions for Checksum Levels - Teradata Database

Teradata Database Design

prodname
Teradata Database
vrm_release
15.00
category
User Guide
featnum
B035-1094-015K

Default Definitions for Checksum Levels

You can use these keywords to specify table‑level checksums for user data tables, join indexes, and hash indexes. You can either redefine their specific meanings using the DBS Control utility or use their Teradata Database‑defined default meanings as defined in the following table.

Note that you must use the default settings for the NONE and ALL keywords. Only the meanings for LOW, MEDIUM, and HIGH are user‑definable.

 

Disk I/O Integrity Checking Level

Checksum Sample Count Percentage

Comments

DEFAULT

Depends

Uses the system‑wide default setting for this type of table as defined by the DBS Control utility.

See “Eligible Table Types for Setting System‑Wide Disk I/O Integrity Checking Levels Using DBS Control” on page 671, and Utilities: Volume 1 (A-K) for more information.

NONE

           0

Checksum integrity checking disabled.

Detects some forms of lost writes.

A lost write is a write of a file system block that received successful status on completion, but the write either never actually occurred or it was written to an incorrect location. As a result, subsequent reads of the same block return the old data.

LOW

           2

Samples one word (2%) of data per disk sector to compute the checksum value.

Detects all forms of lost write and whole disk sector corruption.

Lowest likelihood of detecting bit, byte, or byte string corruption.

MEDIUM

         33

Samples one third (33%) of the data per disk sector to compute the checksum value.

Detects all forms of lost write and whole disk sector corruption.

Higher likelihood of detecting bit, byte, or byte string corruption than LOW sampling level.

HIGH

         67

Samples two thirds (67%) of the data per disk sector to compute the checksum value.

Detects all forms of lost write and whole disk sector corruption.

Higher likelihood of detecting bit, byte, or byte string corruption than MEDIUM sampling level.

ALL

       100

Uses all (100%) data to compute the checksum value.

Detects 99.99999998% of all forms of data corruption.

Strongly recommended for support of the read from fallback feature.

You can specify system‑wide integrity levels at the table type level as well as on an individual base table basis. See “Eligible Table Types for Setting System‑Wide Disk I/O Integrity Checking Levels Using DBS Control” on page 671, and Utilities: Volume 1 (A-K) for more information.