As long as you have the data then you can calculate it. It may be easier with SQL and a normalised form but that forces a lot of other constraints.
With a non-relational DB the idea is to attack it with map reduce and possibly in app calculations. This may not be ideal, but the benefits of easy placement and access of the documents outweigh the aggregation problems on most of the problems I have dealt with.
Neither solution is perfect you just need to make a prediction based on the projects requirements and make a choice.
With a non-relational DB the idea is to attack it with map reduce and possibly in app calculations. This may not be ideal, but the benefits of easy placement and access of the documents outweigh the aggregation problems on most of the problems I have dealt with.
Neither solution is perfect you just need to make a prediction based on the projects requirements and make a choice.