Hello All,

Thanx for the advice u guys.

I also looked at metalink and found out that the problem is
basically due to memory management of large pool which doesn't
use an LRU mechanism.

One way of solving is to increase large pool so did that by increasing from 250 to 300 M.

Will keep monitoring to check if it is enough.

One point to discuss is large_pool_min_alloc

I understand that it is no more required in 8i as large pool
is allocated in chunks of 64 K by default.

The biggest problem is not being able to dynamically alter
sga + large pool in 8i. We will be upgrading to 9i in about
a month but until then are stuck with having to reboot the system / bounce the database to apply init.ora changes.

Physical size (DB Files + Control files + Redo Logs) = 12 GB

Our database is running in MTS configuration with
following 10 dispatchers & 20 servers.

Output from sga is:

Total System Global Area 1095284764 bytes
Fixed Size 75804 bytes
Variable Size 767451136 bytes
Database Buffers 327680000 bytes
Redo Buffers 77824 bytes

Best Regards,

Kashif