We have an AWS-r3.large system. With 1.5TB data volume dedicated to Mongo. We are still using 2.6.3 (yes, upgrade pending for a while now).
The data size is only 550GB. The volume that we have, has 1500 dedicated IOPS. But some how once the IOPS reach 700-800 the Mongo performance, reduces drastically. No clue why are the other IOPS not getting used. Nothing in MMS too. Even if I have a system where I write script that continuously queries on the largest collection in a loop, then too the IOPS go as high as 1200. My largest collection has 150 million documents. Could that be a reason?
Any help would be really appreciated. Lemme know, if anybody requires any more data.