I am following MapR 5.1 Documentation and trying to understand the details related to replication vs NS Replication. I see what's written, but perhaps I am not understanding the "why" to have two different replication settings? Can someone post some examples of situations where an admin may want to have higher or lower values for the replication and NS replication settings? Obviously data loss is bad, so more replication is more better right*? (*Subject to disk space limitations, offer not valid in Rhode Island) Are there any metrics on performance on replication changes? Ie. all else being equal, on a X node (X would have to be more than 6 I would imagine) if I query the same data at 3x replication, and then query it at 6x, what sort of change could I expect? (given a description of node setup and networking etc).
So, other than that, as I alluded to above, I'd also like a specific explanation of replication vs NS... what are teh effects of each, why they are separate, etc. Thanks in advance!
PS Bonus Question - To change the replication of a MapR Table, you change the replication of the volume it resides in... correct?