I spend my day being really satisfied about myself. I made the gamble to lower the memory target for my Oracle database during a data import and it paid off. The first test I had a timing of 5 hours the import, then it was suggested to increase the memory on the machine, because we had some 16Gb to spare on that VM config. So why not give Oracle more memory to play with? Second test timing dropped to 15 hours. I suggested to lower the memory again, but nooo, more memory = faster. I am not a DBA, but I just figured it worked faster with less, so logic says that we test again with lesser memory. So someone else took over, created indices, cleared the recycle bin, cleaned up the audit table and recalculated the statistics. Tests now clocked on 12 hours.
Today during production migration, i gambled and set the memory to the original value, and I clocked the import at 4h58m.
TL;DR: I was right and I am not even the expert on this. Good day.