Browse Prior Art Database

Extended Randomization in Direct Access Storage Device to Lower Hard- Error Probability

IP.com Disclosure Number: IPCOM000117254D
Original Publication Date: 1996-Jan-01
Included in the Prior Art Database: 2005-Mar-31
Document File: 2 page(s) / 60K

Publishing Venue

IBM

Related People

Cunningham, EA: AUTHOR [+2]

Abstract

A method of reducing erroneous data in a Direct Access Storage Device (DASD) from misidentification of track number is described. The method can also reduce the probability of more than one hard error occurring in the same track.

This text was extracted from an ASCII text file.
This is the abbreviated version, containing approximately 52% of the total text.

Extended Randomization in Direct Access Storage Device to Lower Hard-
Error Probability

      A method of reducing erroneous data in a Direct Access Storage
Device (DASD) from misidentification of track number is described.
The method can also reduce the probability of more than one hard
error occurring in the same track.

      DASD designs using magneto-resistive heads, when subjected to
significant vibration or shock while writing, may have significant
erasure of an adjacent track sector, and the data written can be near
the adjacent sector.  It is then possible in attempting to read the
damaged adjacent sector, while using head-shifting recovery
procedures, that the head may be offset toward the encroaching
sector.  The head may then succeed in reading the adjacent tracks
data instead of the track of interest.  Then even when error checking
codes are used, the data is erroneously verified as correct.

      Several methods may be used to prevent the misidentification.
One method is to invert the binary bits of a check byte on every
other track, both while writing and on reading.  This prevents any
data from adjacent tracks from being incorrectly verified.

      A second method is appropriate when a data randomizer is used.
The randomizer starts with a seed, which may have many values.  Thus,
different seeds may be used on different tracks, and the
corresponding seed will be used during the read.  Any data read from
an adjacent track will thus be incorrectly derandomized, and the
check bytes will not verify the...