Key benefits to Nano include smaller VHD sizes and less updates resulting in less reboots and less downtime. From my testing another one of the key benefits is the memory footprint.
For my tests I ran three Windows Server 2016 TP4 systems: One as a Full operating system, one as Core and one as Nano. Each of these were configured as an A1 system in Azure (1 core with 1.75 GB of memory). To check on the amount of free memory, I used the following command in PowerShell:
typeperf “\Memory\Available Bytes”
The results are best shown graphically – when running just the Operating System in each instance:
- The Full version had 25% of its memory available
- The Core version had 50% of its memory available
- The Nano version had 83% of its memory available!
Shown a different way – when running just the Operating System in each instance:
- The Full version was using just under 1.4 GB of the 1.75 GB available
- The Core version was using just under 1 GB of the 1.75 GB available
- The Nano version was using under .4 GB of the 1.75 GB available!
Finally, using the memory which was available in my test (1.75 GB) we could run the following numbers of Full, Core and Nano versions:
- The Full version could run approximately 1.3 operating systems in the memory available
- The Core version could run just under 2 operating systems in the memory available
- The Nano version could run just under 6 operating systems in the memory available!
Summary: The significantly lower memory requirements for Nano (and Core to a lesser degree) make Nano a very interesting choice especially in configurations where memory is a premium and in configurations where it is important to have a higher density of virtual machines. For more details on benefits and how to connect to Nano and Core systems in TP4 see https://www.catapultsystems.com/cfuller/archive/2015/12/03/connecting-server-manager-to-tp4-nano-and-core-servers-running-in-azure/.