Thunderstone Search Appliance Manual

Replication Overview

In replication, a server profile sends walk data to another server profile. The two profiles can be on different machines or they can be on the same one. If the profiles are on different machines, the sending and receiving profiles can have the same or different names. If the profiles are on the same machine, use different profile names.

Here is an example that illustrates the replication process. In this example, the Sender profile has been set up as the sender profile and Receiver is the receiver profile. After Sender performs a walk, it sends the walk data to Receiver. The Receiver profile accepts the data as-is, without regard to its own profile settings. Only the profile that performed the walk may send the walk data, so in this example Receiver cannot replicate (the data it received from Sender) to another profile.

To avoid undesired overwriting of replication walk data, you should not allow the receiver profile to perform walks.

Before the receiver will accept replication data, the sender(s) need to be granted permission to send the data. This permission is managed in a cluster member list.

A good use of replication is to set up multiple machines to replicate to a single receiving profile. For example, machines A, B, and C each have a different profile, and they each replicate their walk data to a profile on machine D, which is the receiver. Another use of replication is to send walk data from multiple profiles on a machine to a single receiver profile that is on the same machine. This provides a means of combining walk data into a single profile. Another use of replication is to replicate data from one sender to multiple receivers. This way multiple machines hold the same walk data.