most operators of RapidMiner need a copy of the data in main memory. For data of this size this is obviously not feasible, so in most cases RapidMiner cannot handle this amount of data at once. But usually you perform some sampling though and work just on parts of the data, and RapidMiner provides a good interface to common databases, so you can the the samples directly from your database. Additioanlly we are currently implementing some algorithms to work directly inside the database, so the data is not at all copied to the RapidMiner memory.
Contributor I
Answers
most operators of RapidMiner need a copy of the data in main memory. For data of this size this is obviously not feasible, so in most cases RapidMiner cannot handle this amount of data at once. But usually you perform some sampling though and work just on parts of the data, and RapidMiner provides a good interface to common databases, so you can the the samples directly from your database. Additioanlly we are currently implementing some algorithms to work directly inside the database, so the data is not at all copied to the RapidMiner memory.
Cheers,
Marius