Browse Prior Art Database

Automatic Resolution of Dependent Objects in a Data System

IP.com Disclosure Number: IPCOM000112970D
Original Publication Date: 1994-Jun-01
Included in the Prior Art Database: 2005-Mar-27
Document File: 4 page(s) / 144K

Publishing Venue

IBM

Related People

Bendert, EJ: AUTHOR [+4]

Abstract

Certain objects in a data system are dependent on other objects, such that typically the dependent object cannot exist until the primary object, on which it depends, is created. Additionally, dependent objects may exist in a separate repository or partition of the data system from the corresponding primary objects.

This text was extracted from an ASCII text file.
This is the abbreviated version, containing approximately 37% of the total text.

Automatic Resolution of Dependent Objects in a Data System

      Certain objects in a data system are dependent on other
objects, such that typically the dependent object cannot exist until
the primary object, on which it depends, is created.  Additionally,
dependent objects may exist in a separate repository or partition of
the data system from the corresponding primary objects.

      Partitioning is introduced for several reasons (ease and
flexibility of managing physical storage...), including flexibility
of doing incremental backup and restores.  It can be seen that
backup/restore of objects in such a partitioned system can encounter
a problem when a dependent object cannot be restored until its
primary object is established.  In the past this problem has been
basically left unresolved.  Dependent objects, whose primary objects
were recreated by the restore process, were very often left in an
unresolved state, i.e., they were not automatically resolved when the
primary object was restored.

The methodology for creation and subsequent resolution of dependent
objects is as follows:

1.  When dependent objects cannot be resolved, the restore process
    leaves dependent objects unresolved (recreates them in a special,
    "unresolved" state).

2.  The restore process creates an entry for the unresolved,
    dependent object in a special systems directory that exists just
    for that purpose.  The entry contains the external name of the
    dependent and primary object.

3.  The restore process keeps a local list of all primary objects
    encountered in the restore of the current partition.

4.  When all object restoration completes for a particular partition,
    there is an automatic search of the special "unresolved,
    dependent objects" system directory for entries that match the
    local primary object list entries.  When a match occurs on the
    primary object external name, the unresolved dependent object
    directory entries are deleted, along with the unresolved objects
    themselves, and they are replaced with fully resolved dependent
    objects.

      The solution is not restricted to the restore application.  It
is generalized to include the function of being able to generally
create dependent objects in the same or different partitions before
the corresponding primary object.  Then it also includes the general
capability for automatic resolution of the dependent object when the
primary object is created.  This is effectively the same as was
described for the restore application.  They work the same because:

o   Any creation of a dependent object that cannot immediately be
    resolved to its primary object causes an "unresolved" dependent
    object to be created and an entry made in the special systems
    directory.  This action is done regardless of whether or not it
    is originated by a restore process.

o   Any primary object creation c...