你问问题不清楚
第一个笨办法是: perform a parallel full table scan in the largest table with a hint like /* full(x) parallel(x, 16)*/. It depends on your server's capacity,you can set different parallel numbers. Then you can joint this table with others by index.
如果你有小一点的table,you can load it into memory by using hash joint. Just check Oracle SQL tuning help.
偶尔路过,帮不了忙。随嘴带两句。BTW, my rate is $250US/Hr.
If you can not handle these tables with records less than millions, better you....