We need a robust statistical software that can handle 10~13 million records datasets as an input to support business decisions, mostly forecasting and multiple regressions analysis. We have already tried R but it couldn't handle such datasets. We are looking for a reliable customer support, regardless of price. Do you have any suggestions?
It would be interesting to see what algorithm in R has failed (and how). I think this question depends on a particular implementation of multiple regressions. I would shop around for a good algorithm. One interesting project that has many implemented algorithms to do regression is DataMelt https://jwork.org/home/datamelt/ It has a good database of examples organized by topics and a search capabilities. Try to put "regression" into its search:
it will return ~10 examples and a large number of implemented algorithms (but all in Java). Personally, I use "polynomial regression" from this tool