I have a question with a distributed deployment...
If a deployment was set-up to have for example:
2 x Indexers
n x Forwarders (set-up to autoLb between the indexers)
1 x Search Head
The autoLB will forward data to the Splunk indexers in a cycle based on time.What happens to the visibility of data if one of the Indexers was to become inactive (e.g. a system failure, etc). I would imagine that Splunk would be able to view ~half of the data, is this assumption correct?
How would data replication between the Indexers take place? - If the there is a requirement for the data to remain 100% visible, what would be best to achieve this?
I'm sure I have come across guidelines on data replication between two indexers in past notes/discussions/Splunk documentation. But I am not able to find the justification I require.
Are there any thoughts on documentation or sources of information that would be useful?
You are correct that if one indexer is out, only half the data will be visible, though the search head will report that it is unable to reach all indexers.
In the current version, there is no native replication of data. You will have to do this either using the underlying storage to replicate, or by forwarding from indexer to a replica instance. Both have disadvantages relative to the other. In addition, there is no built-in mechanism for failover, so you would have to implement this yourself. These solutions are not entirely simple to implement correctly and robustly. An overview of this is here: http://docs.splunk.com/Documentation/Splunk/4.3.2/Installation/Highavailabilityreferencearchitecture
In future versions, you may expect some form of built-in replication, as well as a more automated built-in failover, that should be preferable to these other methods.