TA installation in distributed environment

[chrisdev]

I have a simple TA that makes a request to a REST endpoint and writes the data to the index (no UI associated with this, only indexing).  Im exploring a distributed Splunk environment (with a forwarder, indexer and search head) but im unsure where to install the TA. On the forwarder, the indexer or somewhere else?

Reading similar forum posts it appears that the answer can depend on the TA, however what about a TA decides in what environment it should be installed?

[ephemeric]

It also depends on your network topography. The TA must be able to connect to said REST API.

There is "search time" and "index time" and if your TA does the latter, it belongs on an indexer.

Nothing precludes your TA being installed on a SUF and writing to a file waiting to be forwarded...

[isoutamo · Accepted solution]

That’s true. If TA has search time REST calls or something similar then it should be placed on SH layer. But in those cases, if it doing indexing also it needs to install indexing / forwarder layer too. Personally I tried to avoid installing anything to indexing layer and prefer always dedicated HFs, as almost all changes to TAs cause rebooting or other ways disturbing searches. Only exemptions are indexing definitions which must be on indexing layer.

shortly:

• search time activities-> SH layer
• index definitions -> indexer layer via CM
• indexing activities -> separate HF (usually only into one node)

more https://docs.splunk.com/Documentation/AddOns/released/Overview/Distributedinstall

r. Ismo

[ephemeric]

I like it! Well said.

[isoutamo]

Hi

in normal case on distributed environment all TAs which get data into indexers should installed on (heavy) forwarders. 
r. Ismo