Clustering Message Broker

There are three basic components in MB:

Broker
Cassandra Server
ZooKeeper Server

By default, MB is packaged with all three components. In a clustered setup, it is possible to have these three components separately and shared with each other. This separation introduces several possible deployment scenarios that can be customized according to your requirement.

The deployment pattern of deploying WSO2 Message Broker with external Cassandra Cluster and external ZooKeeper cluster is the production recommendation for WSO2 Message Broker clustered deployment. This deployment pattern provides the flexibility of going for any number of Cassandra nodes, MB nodes and ZooKeeper nodes and it makes it easy to isolate the issues if there are any.

To enable the governance registry to be used across all nodes, you must share the registry. For details, see Sharing the Governance Registry Across the MB Cluster.

Every node where WSO2 Message Broker is deployed (including Cassandra and ZooKeeper server nodes) should be time synced. Please consider as this is very important and strictly recommended as we have met with situations where Message Broker malfunctions without aforesaid requirement. Need administrator permissions:

$sudo apt-get install ntpdate;

$ntpdate pool.ntp.org

Frequently asked questions

What is the Cassandra version to use when clustering Message Broker 2.1.0 with external Cassandra?

Cassandra 1.2.x

What is the recommended hardware configuration for running a Message Broker cluster?

It depends on the number of Message Broker nodes. The Cassandra server should be running on 2 CPU - 6GB Memory (minimum), and Message Broker can run as a 2 Core, 4 GB server.

Can both Business Activity Monitor and Message Broker use the same Cassandra cluster?

Yes. They are in independent key spaces, so they can use the same Cassandra cluster.