If I do a query that is doing a hash join. Is the amount of pga/temp space required based entirely on the table that is hashed (smaller) table. Or will more temp space be required for a larger table? Also, after a table is hashed and the join takes place, is the result set temporarily stored in pga/temp space before being return (or pushed to another operations?). Is there anyway to calculate how much pga/temp space will be required for a hash join if you know how much data will be queried from each table? This could be useful for predictive modelling. For example we have a query that does a hash join in. We know our data sets will increase to X. How much more temp space can we estimate we need for this query?