Windows 10 VDI deployment: Architecture and Taxonomy


Virtual server mirror. Serving Windows 10 pro desktops to 20x thin clients.


Server: 2x Dell r710 12 cores @ 3GHz, 64GB RAM, w/ Intel 2x 10GBE port and dual 850W Power
Storage per server: 2x300GB SSD, 4x146GB SAS 15K, 2x1TB SAS 7.2K
Software: M$ Windows 2012r2 Datacenter or Ubuntu Linux 16.04


Virtual Machines

Per physical server:

  • 1x Windows 2012R2 Domain Controller
  • 1x Windows 2012R2 Application server
  • 1x Bitdefender Appliance
  • 20x Windows 10 Professional (Virtual snapshots)

Required storage space:
+500GB 20x Win10 VMS x 25GB (Much less usually because we will be using virtual snapshots)
+200GB 2x Win2012R2 x 100GB
+100GB Bitdefender virtual machine
+100GB User data
+100GB Main install
= 1TB (Available 1.88TB)

RAID per server

[Fast] 300GB RAID1 – 2x300GB SSD for application server vm,  and bitdefender vm.

[Medium] 392GB RAID10 – 146GBx4 15K for 20x Windows 10 virtual machines, and domain controller vm.

[Slow] 1TB RAID1 – 2x 1TB 7.2K  for backup and share data

8GB USB thumb drive for OS

Filesystem: ZFS


LAN Subnet =

I like to isolate
Static to 100-199(Servers 100-129, Workstations 130-179,
Printer DHCP to 200-239,
and keep 001 as the gateway, 002-099 and 240-254 for utility. ~Doug

Subnet Mask:
Gateway: DNS-1 – DVVMWS01 (DC) DNS-2 – DVVMWS02 (DC) – DVRMLS01 – Host of VM’s – DVRMLS02 – Host of VM’s – 16 – DVVMWW01-16


Server 1 and 2 are directly connected with 10GB connection.
Every other connection is 1GB.
Spare firewall, switch, and thinclients on hand.


1. First 2 characters for the Company,

2. Next two for Machine Type = Virtual Machine or RM for Real Machine

3. Next two Workstation or Server – WS = Windows Server, WW = Windows Workstation, LS = Linux Server, LW = Linux Workstation.

4. The last 2 = nn = number.

5. Example = DVRMLS01 = Domain Vacations, Real Machine, Linux Server, 01



20x HP G4U40UA Flexible Thin CLient T620 AMD GX-415GA 1.5GHz 4GB 8GB SSD ThinPro

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.