Hyper-V - What’s New What’s Cool?
With the Technical Preview 5 recently released, Server 2016 is generally believed to be almost feature-complete. So it’s worth taking a look at the most important new server virtualization innovations with Hyper-V.
With a view to the competition, what’s noticeable is that several long expected features are finally available. One feature sorely missed was for example the ability to hot add and remove network cards and memory. Another useful feature in many environments is the provision of Windows Integration Services available through Windows updates. Up until now snapshots in virtual machines were only supported as standard snapshots. In Server 2016, VSS-assisted application-consistent snapshots will also be fully supported. Obviously the applicability and compatibility with stacked snapshots etc. must still be analyzed. Powershell Direct enables the administration of virtual machines using Powershell on a Hyper-V host independent of network and firewall settings. This is an exciting feature particularly in terms of automation. Discrete Device Assignment also allows direct access to physical hardware (e.g. FC cards) based on SR-IOV virtual machines. The above-mentioned features require the use of Windows 10 or Server 2016 as an operating system.
The virtual machine’s configuration file is no longer saved in XML format but in binary format instead. This is intended to make the environment more robust and also assist faster booting.
As Hyper-V is usually deployed in clusters, several other cluster-relevant improvements have thankfully been introduced. Let’s start with the most pleasing: A rolling in place upgrade is now supported from the Hyper-V cluster deployed on 2012 R2 to Server 2016. The only function which is not supported is an online upgrade of virtual machines to the latest hardware version (currently 7.1) as performed with other hypervisors. This is required to deploy the new features. Failover cluster node fairness is another feature which has been missing up until now and allows virtual machines to be automatically migrated to hosts with enough resources if the VM-running host cannot provide enough CPU or memory resources. This automatic load sharing was up until now only possible with SCVMM dynamic optimization. With TP5, the network stack has been adapted to the network stack in Azure which makes it far easier to set up complex network scenarios.
If the Hyper-V host is rolled out as a nano server, the virtualization platform requires far fewer resources (e.g., 500 MB HDD, for details see http://aka.ms/nanoserver). Streamlining the Hyper-V host has made it more robust and easier to update than a Standard Server 2016 with desktop experience. Server 2016 also delivers even more compute virtualization innovations such as nested virtualization, shielded VMs or Hyper-V Container. We will report on these features soon.