Hi, I am trying to find out an algorithm which could fit the coprocessor architecture to find out pth-percentiles value from the values distributed among the hbase regions. Has anyone worked on something like this and can point me to some algos which can do this?
So the idea is that each coprocessor will do certain calculations and send back to the client and then client can make final touches to find the pth percentile element. Estimation/Approximation algos will work for me too. I tried to find algos, but didnt' find anything which would easily fit distributed values like in hbase. If anyone has worked on something like this before, then it would help me! Thanks and Regards, Mayuresh
