Browse Prior Art Database

Connection grouping in load balancers

IP.com Disclosure Number: IPCOM000236291D
Publication Date: 2014-Apr-17
Document File: 2 page(s) / 38K

Publishing Venue

The IP.com Prior Art Database

Abstract

Connection grouping in load balancers

This text was extracted from a PDF file.
This is the abbreviated version, containing approximately 53% of the total text.

Page 01 of 2

Connection grouping in load balancers

Load balancers can be used in order to distribute workload around a system - so if, say, two connections are requested, specifying the same host and port, the actual connections may be made to different machines, managed by the load balancer.

    This can cause problems when Application Server Facilities (ASF) consumers are being used to connect from an Application Server to a messaging queue manager. In this scenario, an activation specification thread connects to a queue manager and gets the details of messages from specified queue that are suitable for delivery to the Message Driven Beans (MDBs) hosted in the application server The details of these messages are then passed to Server Session threads for consumption. These Server Session threads are then create their own connections to the queue manager in order to get these messages from the queue.

    The Server Session's connection to the queue manager is created based on the connection details of the activation specifications browsing thread. It has been seen that a load balancer can intercept this request and divert this to a different machine - not the one which hosts the original queue manager.

    This can lead either to a rejected connection, or a successful connection to the wrong queue manager, which causes errors when trying to pick up the

messages whose details have been passed by the activation specification.

    There do not seem to be any current methods to resolve this specific problem - load balancers can be configured so that connections are always made to specific machines for a particular client id etc, but these would make all connections from the application server Java Virtual Machine (JVM) to a specific queue manager.

    This invention introduces the idea of connection groups for load balancers. This would mean that instead of connection requests only being distributed around a system based on current workload, round robin etc, it would be possible to specify that a connection request is sure...