Since Stefano has specified that all the datafiles are in the same Disk Group, the idea of possibly improving performance due to using multiple tablespaces is a non-starter. Even using multiple Disk Groups built on top of a S.A.M.E. media configuration will still just devolve to a statmux of the i/o across the available media. Only if the media farm is presented as independent units of i/o that do not compete for the same throughput resources is it possible use tablespace separation to accelerate queries. And even then the situation has to be right. The sizes of available media usually makes this untenable today. But consider this gedanken experiment: Let's pretend we have 7 "luns" constructed from non-interdependent pieces of your SAN presented as 7 separate Disk Groups, and we construct tablespaces T1 though T7. T1 will get the partitions for the most recent 6 Sundays, T2 Mondays, through T7 having Saturdays. Now since Farnham's conjecture that the busyness of data decays with age is likely true, the most recent 7 days will be on independent units of i/o and the most likely queries per day will not compete with each other much, if at all. And, of course, the resulting number of partitions is 42, so we know the answer must be correct. Regards, mwf -----Original Message----- From: oracle-l-bounce@xxxxxxxxxxxxx [mailto:oracle-l-bounce@xxxxxxxxxxxxx] On Behalf Of Leyi Zhang (Kamus) Sent: Friday, May 14, 2010 11:57 AM To: s.cislaghi@xxxxxxxxx Cc: Oracle L Subject: Re: Daily partitioning >The whole table may be about 6Gb <snip> >>Divide into more tablespaces will not do much >>help on query performance on this table, although it's the reason some >>person think tablespace separation will do. -- Kamus <kamusis@xxxxxxxxx> from Stefano: <snip> > Also datafiles are stored in the same DG. <snip> -- //www.freelists.org/webpage/oracle-l