In our organisation we are using NIFI 1.1 I heard some performance issues while reading data from NIFI processor Read eventHUB. I would like to know what are those issues , i would like to contribute and understand how nifi stream based processor operates in terms of closing and opening connection.
Compare to spark streaming job with say near real-time batch job 10 sec ,we are having rdd of all messages at every 10 s . How does in general even hub steaming processor or jms processssor or kafka processor handle connections if they are being triggered with new flow file every 10 seconds. Do they keep the connection open and save those connection in some controller service or it will always a new connection request at every trigger? In nifi cluster mode how does processor handle using multiple consumer approach.
I am trying to fix bug and also want to understand nifi general approach in case of steaming unreliable listeners .