going directly to the question, if we have one compute node with three network HBA each has 4 ports "for example", now I want to distribute the netork workload of the instances running over that compute node like the following case:
network1 "192.168.1.0/24" used by instance 1, 2 and 3 will only use HBA1 ports
network2 "172.25.250.0/24" used by instance 4 and 5 will only use HBA2 ports, and so on.
the idea here is utilizing all HBA network ports and divide them accross different networks so that instances network load is seperated logically and physically.
is this configuration applicable in openstack ?
Hi Amr, I think what you need to have a look at is SR-IOV
This solution, gives you two benefits, increase the performance, cause you wouldn't use virtual NIC's that consume CPU and memory resources.
And do what you want, for example create three VF's, one for each instance 1,2 and 3 on top of HBA1, and another 2 VF's on top of HBA2 for instances 4 and 5.
Don't know if it is what you were looking for. It's covered in the CL310 course in v10.
I hope it helps...