I need some help (or maybe a better angle of attack) connecting a few external users to our Azure tenancy.

NOTE: All resources mentioned below are in the same Azure Subscription, the same Azure Location and the same Azure Resource Group.

There's one virtual network with a Windows 2016 VM running. This v-net (call it MAIN-VNET) has a virtual gateway with a site-to-site VPN configured (call this S2S-VPN). This S2S VPN is configured as a policy-based IPSEC VPN so that the on-premise Billion 7800VDOX can connect - which it does without issue. But, because it's a policy-based configuration, I can't configure a point-to-site VPN in coexistence with this S2S. If it were route-based configuration however, P2S and S2S coexistence would work (and does - I tested it). But the on-premise Billion won't connect because it doesn't support IKEv2 (which is what route-based IPSEC VPN's use). So for now we're forced to use the policy-based config.

I have had to create a new v-net (P2S-VNET), then a new v-net gateway (P2S-VPN) and then configure the point-to-site VPN and client. I've done this and am able to connect to the VPN using the Azure VPN client successfully, but I can't access the Windows server on MAIN-VNET.

I can't create a "vnet-vnet" connection because one of the VPNs is policy-based and this is not supported by Azure.

I've tried creating a peer for both networks to each other (note that I have not ticked Allow forwarded traffic, Allow gateway transit, or Use remote gateways for either), but still cannot access the server.

I've created inbound/outbound rules in the network security group attached to the Windows server that allows ALL ports between the the various subnets each v-net has and this has not helped either.

Does anyone have any suggestions on how I open traffic between the two v-nets? I need it so that any external users who connect through P2S-VPN can access the Windows server on MAIN-VNET. Users on-premise who connect to this Windows server via the Billion device and the S2S VPN have no issues connecting to the Windows server at the moment.

LAN ==> Billion ==> S2S-VPN ==> MAIN-VNET ==> Windows server [OK]

Remote user ==> P2S-VPN ==> P2S-VNET =/= MAIN-VNET ==> Windows server [NOT OK]

2 Answers 2


It seems like you met the requirements for VNET Peering. All you have to do is connect both VNETs and enable Gateway Transit.

Here's an overview:


And here's how to do it:


Just make sure to redownload the P2S client after reconfiguring VNET otherwise you won't have the new routes.

  • Hmmm... I had already peered the two v-nets. They have been "connected" for a day now. I just tried recreating the VPN client, downloading it and connecting; it connects OK, but I cannot access the application our Windows server is running. Is there any additional address spaces or subnets I need to have in each v-net? Or something else I've missed?
    – Reece
    Commented May 25, 2017 at 10:20
  • Did you peered the vnets after or before configuring P2S? As I said, you need to redownload the client. You also have to make sure you've enabled Gateway transit during peering otherwise it will not work. Commented May 25, 2017 at 10:51
Remote user ==> P2S-VPN ==> P2S-VNET =/= MAIN-VNET ==> Windows server [NOT OK]

It is a by design behavior, we can't use Vnet peering in this way(Vnet peering will not route your remote user's network traffic to windows server).

Your goal is to create P2S VPN between on-premise network and Azure Vnet(MAIN-VNET), as a workaround, we can create a windows server in MAIN-VNET, and configure it work as a RRAS VPN server, in this way we don't have to create another Vnet in Azure.

LAN ==> Billion ==> S2S-VPN ==> MAIN-VNET ==> Windows server 
Remote user ==> RRAS VPN ==>MAIN-VNET==>Windows server
  • That was my fall-back. I can install RRAS/IIS on the Windows server being used by our application, but someone mentioned it would be vulnerable to a DDoS. Would I be better to deploy a tiny Windows server instance in the MAIN-VNET and keep the RRAS dial-in SSTP VPN off the production server?
    – Reece
    Commented May 25, 2017 at 10:39
  • yes, I think deploy another VM in MAIN-VNET is better than install RRAS on your windows server 2016.
    – Jason Ye
    Commented May 25, 2017 at 13:34
  • Just checking in to see if the information provided was helpful. Please let me know if you would like further assistance.
    – Jason Ye
    Commented May 26, 2017 at 7:59
  • RRAS is apparently not supported on Windows VM's in Azure - support.microsoft.com/en-us/help/2721672/…
    – Reece
    Commented Jul 3, 2017 at 3:56

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .