Chapter 9. Deploying a New Broker

Abstract

In most large messaging environments there will be multiple brokers deployed. This may be for load management, high availability, or other business reasons. Using standalone brokers this requires manually installing and configuring multiple instances of Red Hat JBoss A-MQ. Using a fabric, however, you can deploy multiple brokers from a single location and easily reuse large portions of the configuration.

9.1. Type of Deployment

When deploying multiple brokers, you need to decide how you want to manage the brokers:
  • as a collection of standalone brokers
  • a fabric of brokers
All of the advanced networking features such as fail over, network of brokers, load balancing, and master/slave are available regardless of how you choose to manage your broker deployment. The difference is in what is required to set up and maintain the deployment.
Using a collection of standalone brokers requires that you install, configure, and maintain each broker separately. If you have three brokers, you will need to manually install Red Hat JBoss A-MQ on three machines and configure each installation separately. This can be cumbersome and error prone particularly when configuring a network of brokers. When issues arise or you need to update your deployment, you will have to make the changes on each machine individually.
If you brokers are deployed into a fabric, you can perform the installation and configuration of all the brokers in the deployment from a central location. In addition, using a fabric simplifies the configuration process and makes it less error prone. Fabric provides tooling for auto-configuring failover clusters, networks of brokers, and master/slave clusters. In addition, it also makes it possible to place all of the common configuration into a single profile that all of the brokers share. When issues arise or you need to update your deployment, having your brokers in a fabric allows you to do incremental roll outs and provides a means for quickly rolling back any changes.