.. _# ha-overview:

Overview
========

Building a highly available distributed system is hard. 60East AMPS
technology greatly simplifies the chore when you use AMPS for message
passing and filtering between nodes in your system. You can configure
AMPS instances to replicate each other, allowing you to build a network
of AMPS instances that will survive both network and system outages.

To enable high availability throughout your entire system, however, your
clients may need to know how to gracefully detect and recover from
outages. This white paper discusses common failure modes and recovery
options for all kinds of AMPS clients, and discusses support available
in the AMPS Client Library.

+---------+------------------------------------------------------------------+
|         | This paper discusses a replication architecture that works with  |
|         | all versions of AMPS that support replication. However, starting |
|         | with AMPS 3.8.0.0, AMPS supports passthrough replication and     |
|         | replication groups, which greatly simplify replication. The      |
|         | instructions in this whitepaper will still work with AMPS        |
|         | 3.8.0.0, but are not the most efficient way to configure         |
|         | replication. Full details on replicaiton in 3.8.0.0 and later    |
|         | are available in the User's Guide for those versions.            |
+---------+------------------------------------------------------------------+