Hadoop Hypercube

Question

Hey, i am starting a hadoop based hypercube with a flexible number of dimensions. Does anybody know any existing approaches for this?

I just found PigOLAPSketch, but there is no code to use it.

Another approach is Zohmg from lastfm, which uses hbase, but seems to be very dead.

I think i will start a pig solution, maybe you have some advices?

Answer 1

This would be very cool/useful. OpenTSDB is an HBase time-series database that might be interesting to look at, they have a clever approach to secondary indexing.

Answer 2

You can also look at gpu based database https://www.kinetica.com/ but this is not open source, requires separate appliances and movement of data from Hadoop to Kinetica infrastructure.