Browse Prior Art Database

Fast Character Value Locking

IP.com Disclosure Number: IPCOM000106427D
Original Publication Date: 1993-Nov-01
Included in the Prior Art Database: 2005-Mar-21
Document File: 2 page(s) / 69K

Publishing Venue

IBM

Related People

Resch, RP: AUTHOR [+3]

Abstract

A method for efficiently locking (potentially) large, variable length string values is disclosed. The string value to be locked is transformed, algorithmically, to a small, fixed length token. An attempt is then made to obtain a lock on the token through some available locking service. It is assumed that a locking service that handles large numbers of short, fixed length tokens is more readably available, cheaper, and will perform better than one which is based on large, variable length strings.

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

Fast Character Value Locking

      A method for efficiently locking (potentially) large, variable
length string values is disclosed.  The string value to be locked is
transformed, algorithmically, to a small, fixed length token.  An
attempt is then made to obtain a lock on the token through some
available locking service.  It is assumed that a locking service that
handles large numbers of short, fixed length tokens is more readably
available, cheaper, and will perform better than one which is based
on large, variable length strings.

      A simple method for locking large, variable length string
values can be used any time the granularity of the lock provided by
the string locking facility can be be relaxed.  In most cases, this
is acceptable.  This is routinely done in a wide variety of data
processing systems.  For example, in many relational database
products, the impression is given to the users of the products that
individual rows or records are locked.  In actual implementation,
many of these products lock a set of rows or records and, although
some added lock conflicts may result (other rows or records in the
same set), the locking functions work as specified.

      If the string locking facility solution can allow some values
that aren't equal to cause lock failures as long as all equal values
cause lock failures, a simple approach can be taken.  To state it
another way, as long as the solution doesn't allow a lock to be
obtained on the value "GEORGE" if a lock on the value "GEORGE" has
already been obtained by another task, then it is acceptable to
indicate that a lock on the value "GEORGE" could not be obtained when
the value it actually conflicted with was the value "This is a very
long string".

      This result can be achieved by implementing a string locking
facility as des...