What would be an appropriate / suitable database sans Oracle that would be suitable for the following conditions?
-
data mining
includes but does not limit to the following:
a. clicking on links
b. capture geolocation data, platform details
c. user behaviours -
there will be a foreseen exponential surge of data being captured as more apps / sites being deployed
-
for now, there are on average 60,000 ~ 80,000 data being captured every 15 minutes for 2 main sites / apps
-
In the near future, we are also planning to release all these data to the various parties involved where they can use it for their own apps / sites.
As of now, we are using mySQL but we found that it becomes very costly as we started to scale up. As such, any advice would be good as a starting point.
Best Answer
Mahout with HBase is probably a good option for very large datasets.
It is designed to run on many cheap commodity servers.
Also, monetDB is an open-source column-wise database which is very fast for data warehouses.