We are currently running a 3 node Hyper-V Cluster on Windows Server 2016 Datacenter. The VMs running on our cluster are using a total of 542 cores (Hyper Threaded) however there are only 352 cores available in total on the 3 nodes of the cluster. It is possible to achieve successful over-subscription such as this using the Hyper-V classic scheduler, which has been the default for all version of Windows Hyper-V including Windows Server 2016. Server 2016 now provides a new alternative called “the core scheduler” Details about the different types of schedulers is explained here: https://docs.microsoft.com/en-us/windows-server/virtualization/hyper-v/manage/manage-hyper-v-scheduler-types To use the new core scheduler, we would need to have enough hyper threaded cores available on the nodes of the cluster to cover the 542 cores currently needed by our VMs plus enough to allow for growth and to allow for a node of the cluster to go offline. We are considering buying more nodes and going to the core scheduler because it is recommended here https://support.microsoft.com/en-us/help/4457951/windows-server-guidance-to-protect-against-l1-terminal-fault to add an extra level to guard against speculative executions side-channel vulnerabilities. Also, in the first link it says, “The core scheduler will be used by default starting in Windows Server 2019. On Windows Server 2016, the core scheduler is optional and must be explicitly enabled by the Hyper-V host administrator, and the classic scheduler is the default.” We will want to upgrade our Hyper-V Cluster to 2019 at some point and the way it sounds we will be required to use core scheduler to migrate to 2019. So my question is, is it possible to migrate to Server 2019 and still use classic scheduler by changing the setting in 2019 until we have enough nodes with enough cores to support core scheduler or is that not an option?
↧