Browse Prior Art Database

Error Recovery for Magneto-optical Disks

IP.com Disclosure Number: IPCOM000099508D
Original Publication Date: 1990-Jan-01
Included in the Prior Art Database: 2005-Mar-15
Document File: 4 page(s) / 151K

Publishing Venue

IBM

Related People

Friedl, TJ: AUTHOR [+4]

Abstract

Magneto-optical devices require an erase pass and a write pass to overwrite data. The technique described here minimizes the chance of data loss caused by a failure after erasing but before writing.

This text was extracted from an ASCII text file.
This is the abbreviated version, containing approximately 45% of the total text.

Error Recovery for Magneto-optical Disks

       Magneto-optical devices require an erase pass and a write
pass to overwrite data.  The technique described here minimizes the
chance of  data loss caused by a failure after erasing but before
writing.

      The technique developed to deal with updates on the
magneto-optical media first attempts to identify the risk associated
with various write operations.  In the environment where this device
will be used, write operations can effectively be grouped into three
classes:
    -    Sequential data writes
    -    Random directory writes
    -    Random data writes
    When writing data sequentially to a file, the loss of the last
few write operations is not likely to be significant.  If writing was
not completed, it is not obvious that the possible loss of another
physical sector will really make any difference.  Therefore, there is
no attempt to protect writing sequential data.

      Directory updates are also simple to classify.  The loss of the
directory could make it impossible to access any data on the media.
Therefore, directory updates must be protected.

      Random data updates or writes are less clear.  Some data should
probably be protected, while in other cases, the ability to detect a
loss of data would be sufficient.  For these reasons, the option for
protection and a performance degradation will be made available to
the user for this case.

      The protection technique itself involves duplicating the data
which is to be written to the sector being erased. However,
duplication on the magneto-optical media would be an unacceptable
performance degradation.  Duplication on the media would mean that
four revolutions of the media would be required for each "single"
write which was protected.  An alternative exists which provides
protection which is nearly as good, with much better performance.
The alternative implementation is to duplicate the data on another
device--in this case, the magnetic disks used for normal paging
operations.

      The technique begins with the allocation of a "permanent" space
when the I/O manager initializes.  This space is a part of the
virtual address space which will continue to exit until explicitly
destroyed.  When the I/O manager for the magneto-optical device
initializes, it checks for the existence of this recording space and
creates it if necessary.

      Before going further, it is necessary to mention a convention
established for this media.  The "volume id" consists of two parts:
a user-supplied volume id and a system-generated id field.  This
means that even if multiple volumes are given the same user id, the
system- generated id could uniquely identify a particular volume.

      Each time a new media is placed in the drive, the volume id
(both the user and machine-generated parts) of the media is recorded
in the work area.  As operations continue to the magneto-optical
med...