Resource Manager manages CPU usage by controlling the database load to a very precise level.
By default, this level is set to the number of CPUs. That means that on a 4 CPU server,
Resource Manager ensures that no more than 4 Oracle processes (more specifically, foregrounds
and non-critical, CPU-intensive backgrounds) are running at a time. By controlling the database
load, critical backgrounds are able to run in a timely manner and the load on the server is
The above quote is from article: http://www.oracle.com/technetwork/datab ... 133705.pdf
My question is, how does ODRM handle a shared CPU environment among multiple LPARS? If the same workload (90 active sessions requesting cpu) on 100 CPUs the system is only 90% utilized (relatively good utilization) and ODRM is not kicked in. However the system could see that a different LPAR needs CPUs out of the pool. Now the same workload (90 active sessions requesting cpu) is running on 80 CPUS. This scenario has the system theoretically at 112% utilized (90 active sessions / 80 CPUs). Can ODRM recognize this as a fully utilized system or is this a system loophole that ODRM can not help with?
i am currently running Oracle 11.2.03.3. IBM 6.1 power 7 AIX 64 bit. With multiple LPARs with variable CPUs