I wanted to know how riak handles two queries when the data set starts getting really large. Since riak buckets are just key space, everything seems like it lives in one massive index well bucket. I am finding it hard to plan for optimizing my queries as my data grows, keeping slow data away from fast data.
What is the best way to get keys into map reduce, because doing map reduce on a bucket may only enter the items from the bucket? Does it still go though all the keys in all buckets because the bucket is just part of the name of the key?
Is riak map reduce designed to be run in production for web page loads, or is it only really for taking aggregates of data and waiting just like I would with hadoop?
Edit
It is kinda what I am looking for, but what i mean is lets say i work with a lot of links how does it keep the link index fast, are all the links in the same index or different index Riak seems like its just buckets of buckets of information and that even if i have two different link type or links groups or what ever they will be in the same index meaning over time they will get slower even if one link type i keep it small for a reason
