Microsoft Ignite 2019 has seen many new announcements, and the HyperV Team were keen not to be left out. A mix of features for both Azure HyperV based VM’s and on-prem physical HyperV nodes/clusters were included.
The team were very happy to announce that Gen2 HyperV VM’s are now available in Azure offerings, a number of advancements in the hyperv ecosystem meant that gen2 support was a critical requirement. In Azure, VM’s on pakcages such as Mv2 (massive scale), NVv4 (GPU enabled VMs), HBv2 (compute based VM’s) and LSv2 (Storage Optimized) all required Gen2 support to progress to where they are today.
Building off of Gen2 support, VM packages such as Mv2 can now be deployed with up to 416vCPU Cores and 12Tb of vRAM, numbers thata re not yet available in other cloud offerings. HBv2 VM’s, the compute operation optimized ones, can now make use of up to 80k compute nodes in a single instance!
You heard that right, the NVv4 package offers GPU accelerated environments in Azure, and this technology is also filtering down to on-prem/dedicated HyperV Nodes. The HyperV Team have worked to bring a new form of GPU Partitioning to HyperV, a single GPU can be partitioned and chunks of it can be presented to single VM’s as a complete physical pass-thru GPU. The Vm knows no different, and can use the GPU as if it was physical. This is exciting for the accelerated VDI market, and is sure to make waves after the shortcomings of RemoteFX.
A small gesture, but the HyperV Team recognise that application debugging around the HyperV stack hsa often been a tiresome chore. In an effort to make it more palettable, they have released the HyperV stack debugging symbols packages, so analyzing core and memory dumps in crashes etc is now a lot more meaningful.
Windows Admin Center is swiftly becoming the crown jewel in Windows Server 2019+ environments. To that end, the HyperV Team have incorporated Live Migraton control and feedback between HyperV Nodes in Cluster right into Admin Center. This is available today, and in their roadmap, the plan is to eventually use Admin Center as the one-stop-shop for HyperV Sysadmin operation.
Available now in HyperV is the ability group CPU’s in multi socketed physical environments. This presents a lot of optoins around segmenting high risk or public facing VM’s to one CPU set, and high data value VM’s to another. This is in an effort to further mitigate the impact of being vulnerable to CPU Microcode attacks such a Spectre and Meltdown. CPU groups have no knowledge of the other CPU groups, and comms between groups is prohibited.
All in all, exciting times to be involved in HyperV rollouts, both in Azure and on-prem/dedicated.