Hi Ghassan
Currently we only use roughly 45% CPU per node in peak time (we have 22
Cores per EXA database node) but the runqueue is around 40 due to high
number of processes. If we move to Exadata CC I guess during peak time (4
to 6 hours per day) I must scale to 22 core per node at minimum otherwise I
will have CPU contention. Is the scaling automatic and dynamic? How does
the database deal with the change of CPU_COUNT? I also have some unexpected
peak time in different time of the day (sometimes the business users run
batch without notifying the database team), how Exadata CC deal with such
situations?
TIA
On Tue, Jan 21, 2020 at 6:26 PM Ghassan Salem <salem.ghassan@xxxxxxxxx>
wrote:
The enabled CPUs are charged, used or not. BUT, you can enable a small
number, and scale-up/down as you need. If, for example, you went from 20
ocpus to 30 for 2 hours, then back to 20, you'll be charged 10 ocpus for 2
hours.
On Tue, Jan 21, 2020 at 6:20 PM Ls Cheng <exriscer@xxxxxxxxx> wrote:
Hi
Has anyone got practical experience with Exadata CC and know how they
charge OCPU?
Does Oracle charge per number of enabled OCPU or the CPU cycles used? In
the second case for example there are 40 enabled OCPU but if the usage is
average 20% per month only 20 OCPU is charged?
TIA
