Archives of the TeradataForum

Message Posted: Fri, 22 Oct 2004 @ 15:50:29 GMT


	<Prev	Next>		<<First	<Prev	Next>	Last>>

Subj:		Re: V2R5 Upgrade and CHECKSUM

From:		Morris, Mark

Mark >> the checksum may be applied to some or all of the data corresponding to each disk sector.

Victor > I recently had a chance to look at some CRC calculations (outside of Teradata but that's aside) and am now convinced that it is not a luxury thing to calculate them on a regular basis. Thanks for the heads up on this new feature in TD. Could you explore a bit more on the last part you mentioned, i.e., the disk sector level?

The Teradata Disk I/O Integrity Check Feature generates a checksum by using an XOR based algorithm with the data inside a data block used as input before writing the data block to disk, and stores the checksum external to the data block being written. When the data block is read back from disk, the checksum is calculated again and compared with the previously generated checksum. Access to the data block is only provided to the database if the two checksums match. A sampled statistical checksum algorithm where all or only a portion of the data in each disk sector of a data block is XOR'd together is used to calculate a checksum. The amount of data to sample is user tunable; the checksum can be calculated by sampling a single word per disk sector, multiple words per disk sector, or the entire data in the disk sector.

Here is some information extracted from the user documentation.

DBS Control Record - Disk I/O Integrity Fields:

     CHECKSUM LEVELS

     1.  System Tables            = NONE
     2.  System Journal Tables    = NONE
     3.  System Logging Tables    = NONE
     4.  User Tables              = NONE
     5.  Permanent Journal Tables = NONE
     6.  Temporary Tables         = NONE

     CHECKSUM LEVEL DEFINITIONS

         NONE                     =   0% Sampling
     7.  LOW                      =   2% Sampling
     8.  MEDIUM                   =  33% Sampling
     9.  HIGH                     =  67% Sampling
         ALL                      = 100% Sampling

DBSCONTROL Checksum Screen

Checksum sampling levels can also be specified on an individual table basis with the CREATE TABLE, CREATE JOIN INDEX, and CREATE HASH INDEX SQL syntax. The CHECKSUM option can be specified with any of these commands, and has the follow-ing syntax:

     CHECKSUM = DEFAULT | NONE | LOW | MEDIUM | HIGH | ALL

A DEFAULT setting means to use the current checksum setting based on the table's table type specified in the checksum screen of the DBSCONTROL utility. If no CHECKSUM option is specified with a CREATE statement, then the DEFAULT setting will be used.

For example, a CREATE statement which specifies the CHECKSUM option looks like:

     CREATE TABLE V, CHECKSUM = LOW (a INT, b INT);

The ALTER TABLE command also has a similar checksum option in its syntax:

     CHECKSUM = DEFAULT | NONE | LOW | MEDIUM | HIGH | ALL [IMMEDIATE]
     ALTER  TABLE V, CHECKSUM = MEDIUM IMMEDIATE;0


	<Prev	Next>		<<First	<Prev	Next>	Last>>

Attachments

Library

Quick Reference