Browse Prior Art Database

Directory Maintenance in Distributed Processing Systems

IP.com Disclosure Number: IPCOM000123173D
Original Publication Date: 1998-Jun-01
Included in the Prior Art Database: 2005-Apr-04
Document File: 1 page(s) / 62K

Publishing Venue

IBM

Related People

Banks, AD: AUTHOR [+3]

Abstract

In a distributed computing system consisting of a number of computers a single directory is required to retain knowledge of the current location of resources. This directory (or at least copies of the relevant parts) must be available to all of the computers. Several approaches to this problem are already available, for example the DCE directory service or the internet Domain Name Service. These have drawbacks, particularly in the areas of availability of the directory and its consistency when network links are unavailable. These drawbacks are even more severe where asynchronous processing is involved, as is the case with messaging middleware such as MQSeries*.

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

Directory Maintenance in Distributed Processing Systems

   In a distributed computing system consisting of a number
of computers a single directory is required to retain knowledge of
the current location of resources.  This directory (or at least
copies of the relevant parts) must be available to all of the
computers.  Several approaches to this problem are already
available, for example the DCE directory service or the internet
Domain Name Service.  These have drawbacks, particularly in the
areas of availability of the directory and its consistency when
network links are unavailable.  These drawbacks are even more severe
where asynchronous processing is involved, as is the case with
messaging middleware such as MQSeries*.  These drawbacks are overcome
as described here by the use of publish subscribe paradigms to
distribute directory information to interested parties (subscribers)
ahead of its intended use.  This allows the directory access to be
made at the time of use, even though the network may be unavailable.
This is made possible by exploitation of the rule that only the owner
of a resource may control the directory entry representing the
resource.  Even though there may be multiple copies of the directory
available for availability reasons, there is a single source of
updates for each entry in the directory and hence an unambiguous
order for applying the updates to all of the copies of the data.  A
typical implementation is in the context of saving information about
the location of MQSeries queues in a group of MQSeries queue
managers.  In addition to the queue managers there are a few
computers that act as directory brokers....