High availability planning

High availability (HA) enables messaging services to continue over the outage of a server. In an HA environment, two servers are connected and configured as an HA pair so that, if one of the servers (the primary) fails due to a planned or unplanned outage, the other server (the standby) can take over as the primary and messages can continue to be processed. When the original primary becomes available again after the outage, it acts as the standby in the HA pair.

Application requirements

Your applications can benefit from HA. They can connect to the standby node in the event of a failure of the primary node in the HA pair. They must:
  • Specify a list of endpoints for the Eclipse Amlen servers that make up the HA pair.
    • In MQTT, you can specify a list of server URIs.
    • In JMS, you can specify a list of servers by using the Server ConnectionFactory property.
  • Attempt to connect again after a failure.
.
For more information about configuring HA applications, see the following topics:
For more information about configuring HA with external resources, see the following topics:

Virtual Edition

On a SoftLayer® virtual server instance (VSI) and a SoftLayer bare metal instance (BMI), the following two network interfaces are available:
  • a private interface on a network in the IP address range 10.0.0.0 - 10.255.255.255
  • a public interface
On an Amazon EC2 instance, the following network interfaces are available:
  • a private interface where you must attach an Amazon elastic public IP address to serve as a public interface
  • a private interface

In an HA environment, there are two network interfaces for each server - a replication interface and a discovery interface. When you are planning your configuration, use the private interface for the replication network interface controller (NIC) and use the public interface for the discovery NIC.

For more information about high availability, see Configuring your system for high availability.