Browse Prior Art Database

Server Isolation in Message Queueing System

IP.com Disclosure Number: IPCOM000117137D
Original Publication Date: 1995-Dec-01
Included in the Prior Art Database: 2005-Mar-31
Document File: 2 page(s) / 76K

Publishing Venue

IBM

Related People

Carter, JM: AUTHOR [+3]

Abstract

Message Queueing systems transfer data from one application (the client) to another application (the server) by way of a (potentially multi-routed) network, by use of an intermediate entity (the queue) in order that the client and the server can operate independently with respect to time. The server and the client perceive one (and only one) queue.

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

Server Isolation in Message Queueing System

      Message Queueing systems transfer data from one application
(the client) to another application (the server) by way of a
(potentially multi-routed) network, by use of an intermediate entity
(the queue) in order that the client and the server can operate
independently with respect to time.  The server and the client
perceive one (and only one) queue.

      The general case is one in which multiple client applications
are queueing multiple messages to the same queue, which can be
processed by multiple server applications.  The messages queued by a
single client may all be within the same unit of work, or within
different units of work.  The clients must have the capability to be
independent (autonomous) of each other.  Similarly and separately,
the servers must have the capability to be independent (autonomous)
of each other.

      Multiple messages from one client to the same queue (within a
single transaction or over several transactions) lead to expectations
that only one of the possible servers will process all of those
messages; and also to expectations that the server can receive the
messages in the order in which they were queued.

      In practice, the possibility of the different messages
navigating the network by different routes leads to the potential
that the messages may arrive at the queue in a sequence different to
that in which they were sent.  This problem is easily solved by the
use of (unique) sequence numbers.  But this does not prevent two
independent servers from interfering with each other, for example, by
server A getting the first message and server B getting the next (and
so on).

      The technique described here uses the concept of a "group of
messages", so that the client application can indicate to the message
queueing system that certain messages are associated into a
"collection" (group).

 ...