This is the twentieth part of the Availability Anywhere series. For your convenience you can find other parts in the table of contents in Part 1 – Connecting to SSH tunnel automatically in Windows
We already know how to use full-tunnel VPN from a virtual machine on the host with TCP over File System. Today we’re going to see how to nest full-tunnel VPN inside another full-tunnel VPN.
Imagine that you take your corporate laptop and go to the office. You want to connect to your client’s network, so you use full-tunnel VPN (something like Cisco AnyConnect or Wireguard). Your client configured their firewall to allow your office’s IP address. All works.
Now, you want to work from home. You take your corporate laptop and come back home. You try connecting to your client’s network, but they didn’t allow your home’s IP address to connect. You try connecting to your office’s network with a VPN and this works well. However, you can’t connect to your client’s VPN now because you can’t have two full-tunnel VPNs enabled. How to deal with that?
I managed to get something like that to work with the following setup:
- On the host, install Hyper-V
- On the host, install Hyper-V guest called VM_A. I used Windows 10 x86. It’s better to use x86 VM to avoid some issues with nested virtualization.
- On the host, enable virtualization extension for VM_A with
Get-VM | where Name -eq "VM_A" | Set-VMProcessor -ExposeVirtualizationExtensions $true
- On the host, connect to Hyper-V guest VM_A with shared drive
- In Hyper-V guest VM_A, install Virtual Box 4.2.36 which works on Windows x86. Some other version of Virtual Box possibly can work.
- In Hyper-V guest VM_A, create Virtual Box guest VM_B. I used Windows 7 x86. Again, use x86 VM version.
- Configure VirtualBox guest VM_B networking as NAT with default interface type (something like Intel PRO/1000 MT Desktop)
- In VirtualBox guest VM_B, install Virtual Box guest addins
- In VirtualBox, configure shared directory
\\tsclient\Cwith full read-write options (uncheck read only) and with automount
- In Hyper-V guest VM_A, configure VPN to your office’s network. Let’s call this connection VPN_A.
- In VirtualBox guest VM_B, configure VPN to your client’s network. Let’s call this connection VPN_B
- You may need to set DNS in VirtualBox guest VM_B to something like
- Enable VPN_A, then enable VPN_B.
All should work. Mind that in order to use VPN based on layer 3 GRE protocol (like PPTP) for VPN_A, you may need to reconfigure Hyper-V guest to use bridging. I don’t know if it’s possible to use GRE-based VPN for VPN_B in this setup (I doubt that). However, typical corporate VPN-s use HTTPS on the way to avoid issues.
I tested this setup with Wireguard as VPN_A and Windows’ SSTP as VPN_B. It worked correctly. When VPN_A was enabled and VPN_B was disabled, then VirtualBox guest VM_B was visible to the internet from the IP address of VPN_A (so the IP address of the office). When both VPN_A and VPN_B were enabled, VirtualBox guest VM_B was visible to the internet from the IP address of VPN_B (client’s IP address). Also, shared drive from host was visible in VirtualBox guest VM_B when both VPN_A and VPN_B were enabled. Therefore, you can use TCP over File System properly to route the traffic from the host through any of the VPNs.