For those of you with large deployments, how big are your indexer clusters and what are the limits?

[jeffrogers]

I'm very curious. For those of you with large installs, how big are your clusters? Where are the limits? If greater than 10 nodes, how many of your cluster nodes are receiving data from your heavy forwarders? Do you send data to be indexed to all of them? A subset of them?

[mataharry]

I work on a cluster with 5 indexers with replication.
- 5 indexers + 1 Search-head + 1 Cluster-master, all on linux and splunk 6.2.4
- the replication factor is at least 3, search factor at least 2.
- - So we can lose 1 indexer without search impact, and we can loose up to 3 indexers without data missing.
- it it's fine for 500GB /day of ingestion. (~ 100Gb/day per indexers is a safe ratio with the replication)

I do not use Heavy forwarders in the mix, only Universal forwarders with splunktcp-SSL. The event parsing is done by the indexers
So far it works well for up to 2000 FWD per indexers.