We are using custom java program to index data to splunk using Java SDK (126.96.36.199). We have huge volume to data to index on daily bases around aprox.100000 documents.
We used TCP method but it requires a port to open and bind with a specific source to index data. We receive data as bulk (several records) and process it and index each record seperately. to index data as unique event we need to connect TCP port each time and close it to index data. This way it is working very slow (might be due to connecting and closing TCP port for each record). We tried the com.splunk.Index API (submit()/ attach()), but index rate is very slow.
We have java parsers that are doing some business specific stuff with data, so dont want to go for forwards to rewrite the data parsing logic.
Can anyone help here and suggest the best practice/approach to index, using java SDK (considering high data velicity).