That suggests that you have something like an intensively accessed sequence that is not cached or has the ORDER clause specified, or some other block that is modified repeatedly and intensively.

I used DBMS_SUPPORT package to trace the session with a value true for both waits, and binds. The p3 values are either 1016 (60% of the time) or 1012.

If the trace you did was a 10046 trace at level 8, then what were the p3 values for the buffer busy waits?

I have an oracle 8.0.5 database running on HP UX11.0. v$waitstat shows very high waits on data block, undo header, undo block. I checked x$kcbfwait to determine, which files are affected most. I reviewed all the pctfree/pctused for all the tables, bumped INITRANS to 9, freelists 3 for all index and tables (after dropping and recreating), changed tranasaction_per_rollback_segs to 1. But it still doesn't seem to take care of my problem. I did a trace on the process and found Buffer busy waits on the table/index that i changed. What more should i do to eliminate these waits ?