Not able to increase Hbase read throughput

I have 50 GB data in hbase. I only have one column in table. My throughput is not increasing from 500 rps. I need at-least 5000 rps. I tried changing various parameters but it is not working. I found suggestions on internet which says to manually split records in your regions I am not able to understand how to manually split. Can anyone explain in detail about the process of manually splitting data in regions.

    ["-s","hbase.hregion.max.filesize=1073741824",
"-s","hfile.block.cache.size=0.6","-s","hfile.regionserver.global.memstore.size=0.2",
"-s","hbase.client.scanner.caching=10000",
"-s","hbase.regionserver.handler.count=64"]

Another problem I have is while making changes in the the xml if I am not able to restart the cluster. It throws the error of public key denied , I think It is because master tries to connect every core node which requires the pem file. Any idea how to restart the cluster.