Hi all, I've been working on fixing/refactoring VMware portgroup reconfigure on network offering change, https://github.com/apache/cloudstack/pull/5181 While working on it I found different portgroups on the same VLAN can be used based on the different network rates used in the network offering of the network and compute offering of the user VMs. Below are my findings:
* Network offerings will control network rate only for VR(s) and non-default NICs of the user VMs. * Default network rate for VRs and non-default NICs of the user VMs can be controlled by global setting, "network.throttling.rate". * For the default NIC of user VMs, the network rate is controlled by the compute offerings of the VM. * For the default NIC of user VMs, the default network rate can be controlled using the global setting,"vm.network.throttling.rate". So now my queries are: * Is having VM-level (using compute offering and global setting for VMs) useful? Having a higher network rate for VMs and a lower network rate for VRs(network) will make it useless. * Should there be a single portgroup in vCenter based on the network rate of VRs/network especially in the case when the network rate for VRs is lower? * Even while using different network rates, should there be consistency across different NICs of user VMs? Currently, compute offering and global setting only affects the default NIC of the user VMs. Please share your thoughts and ideas. It will be interesting to know how others configure network throttling for networks and VMs. Regards, Abhishek