> I am looking at activating "shared servers" in one of our 9i instances. > I am looking at an old document I have from when did this in 8i in 2004. > At that time, we decreased the size of the shared_pool when we created > the large pool. > > I can not find anything on ML to explain why I would have done this? > Do normal non-dedicated connections also use the large pool instead of > the shared_pool for some storage? Any other ideas why the shared pool > would have been cut in size? > > > Jeffrey Beckstrom > Database Administrator > Greater Cleveland Regional Transit Authority > 1240 W. 6th Street > Cleveland, Ohio 44113 Could that be related to the fact that before 10g, a server process in shared server config uses UGA for sorting and hashing and UGA comes from large pool (unless large pool is not configured)? See http://yong321.freeshell.org/oranotes/LargePoolMtsPga.txt Since the physical memory of a box is fixed, when you create large pool and if you don't want to increase SGA, you can shrink shared pool or buffer cache accordingly. But I think there's advice that you go the other way, i.e. *increase* shared pool when you run shared server config. Yong Huang -- //www.freelists.org/webpage/oracle-l