buffwaits

The buffwaits column of onstat -p on our system seems to have a high percentage of users that are waiting for a buffer. I have doubled the LRUS from 8 to 16 and BUFFERS from 2800 to 5600 in onconfig and there has been no difference in the ratio of bufwaits to bufreads/buffwrites. Am I on the wrong track here? We have a OLTP system with one table in particular that has large numbers of insert functions applied ot it. Is it possible that the bufwaits are partially caused by several users wanting to simultaneously insert into this table since it is but one page in the buffer pool? Any input on the question, however brief, would be appreciated. Configuration: IBM RS6000 4 X 166mhz processors, 1GB ram, IDS 7.30 UC5, 2GB database on four 2GB drives. ==================================================================== Harold Luse Phone: (970) 491-4120 Veterinary Teaching Hospital Fax: (970) 491-4123 Colorado State University Pager: (970) 229-8173 Fort Collins, Colorado USA E-mail: hluse@vth.colostate.edu ====================================================================

Luse Harold wrote: > > The buffwaits column of onstat -p on our system seems to have a high > percentage of users that are waiting for a buffer. I have doubled the > LRUS from 8 to 16 and BUFFERS from 2800 to 5600 in onconfig and > there has been no difference in the ratio of bufwaits to > bufreads/buffwrites. Am I on the wrong track here? > > We have a OLTP system with one table in particular that has large numbers > of insert functions applied ot it. Is it possible that the bufwaits > are partially caused by several users wanting to simultaneously insert > into this table since it is but one page in the buffer pool? > > Any input on the question, however brief, would be > appreciated. > > Configuration: > IBM RS6000 4 X 166mhz processors, 1GB ram, IDS 7.30 UC5, 2GB database on > four 2GB drives. Post that onstat -p output and we will try to advise. Art S. Kagel