The following suggestions may help you:
1 Increase redo log file and log buffer size.
2 Ensure that checkpoint occurs once in 15 minutes.
3 Keep minimum number of indexes on the table.
4 Allocate big extent size for this table
5 Increase DML_LOCKS parameter in INIT.ORA file and bounce the database.
6 If the table uses SEQUENCE, increase the cache size.
7 If hardware level log files are mirrored as well as software level on RAID 1+0 disks, drop those files. Create the mirrored log files on non-mirrored disk.

PS. Parallel Server is suitable for distributing the application load to many CPUs, not suitable inserting rows in the same table. If 300 users are connecting into 2 machines and accesing the same table, there could be many waits because of the lock on data block. You can experiment by connecting all users through one CPU and collect the wait statistics, and compare this statistics with the statistics collected from 2 CPU (more).


[Edited by tamilselvan on 12-14-2000 at 03:03 PM]