About capacity management and fault tolerance implementation

This section introduces common methods for managing capacity and implementing fault tolerance. Subsequent sections provide more details as well as example implementations of these methods in your message architecture.
A common way to add capacity and fault tolerance to your architecture is to create or expand clusters. Clusters are sets of load sharing systems that perform the same step in your message architecture. You can cluster multiple MTAs and multiple
Network Prevent for Email
Servers. You can increase the number of
Network Prevent for Email
Servers independently of the number of MTAs.
Some ways to add capacity and fault tolerance to your architecture are as follows:
  • MX-based clusters are useful for managing capacity and implementing fault tolerance. You can use mail exchange records (MX records) to distribute email services over a cluster of message handlers. This method of clustering works both for MTAs and
    Network Prevent for Email
    Servers. To create this kind of cluster, each message handler must have SMTP listeners on the same TCP ports. Each message handler must also process inbound messages in exactly the same way. The MX record for each message handler in the cluster must have equal MX preference values.
  • IP load balancer-based clusters are useful for managing capacity. IP load balancers devices provide a virtual IP address that distributes traffic among several back-end servers over an internal IP network.
  • MTA-based queue management is useful for implementing fault tolerance. One or more MTAs in a cluster may be able to check the age of messages queued for a
    Network Prevent for Email
    Server. If any messages have been in the queue longer than the configured limit, such MTAs move them to a queue for a different message handler.