Need help with remote access solution

The correct link for the drawing:

I can imagine that clients are not pleased with a VPN. Anyone that has VPN access will basically have an external computer attached to their internal LAN. This can impose major security problems and I would never recommend it.

There are solutions where a computer at the client has an outgoing connection to the internet. Across that connection you can establish connections backwards to the customers network to internal devices.

We use this to perform remote support to servers, workstations, printers and all other networked devices at our customers. We even have solutions where both sides have no incoming connections at all and we do not know each others IP address! You can find a brief description of this solution here:

This method could be used in your situation, but you need a computer at the customer that has outgoing internet access (this is your ?????). No computing power is required, so it could be a small box that doesn't even looks like a computer. We use fan-less mini computers that draws only 15 to

20W of power. This computer acts as a router for all your communication protocols and provides strong encryption accoss the internet.

What communication protocols do you need to access systems at the client? For how many customers do you need this remote access?

If the number of customers is low, youou could install all remote access software on a computer at the client and provide remote access to this computer. This requires a more expensive computer system and may cause licensing problems, while every software installation may require a license for the remote control software.

Frank

Nice solution, but what makes it different (from a security point of view) from having a VPN connection into the client's network? The support workstation still has full access to anything on the client's network, only limited by what the intermediate server allows. And that server is outside the client's control.

What we normally do is declare all systems that have to be remotely accessible for maintenance a separate DMZ, and have a firewall (under client's control) between that and the rest of the network. Only what is absolutely needed for operations is allowed through that firewall.

Remote access can be via a modem, via a VPN into the DMZ, or by any other means. Most IT departments are quite comfortable with that; some of the more paranoid ones have the incoming VPN connection set up so that only certain IP addresses are reachable through the VPN.

j.

The Remote Support Server/Client have to be configured for each individual protocol and device that needs to be accessable. To create a connection the helpdesk needs a special application with username/password for setting up a connection with the Server and being switched to the Remote Support Client. This is the only way to make connections. No other devices in the customers network are accessable, which is the case with a VPN. The customer has an overview of all connections that the installed Remote Support Client supports.

VPN connections can be restricted via DMZ, but most IT departments and even VPN suppliers have problems implementing it properly. In larger companies it is very difficult to get any approval to modify a router, let alone a firewall, so we cannot touch existing security systems. I have experiences where VPN and firewall suppliers did not believe our system works, because their security was so good. How ignorant and dumb can they be?

This system requires no computer, network or firewall modifications. Just a black-box at the customer and an 1MB application at the helpdesk. It is not only used for remote support applications, but also for remote access like Microsoft's Remote Desktop (RDP). Several agencies use this to add an extra layers of security upon RDP: extra encryption and compression, extra authorization layer with access restrictions and most appreciated: Windows Servers need no incoming internet connections so they are less vunerable to attacks.

Our application has extensive security measures located at the Remote Support Server. They are not based on IP addresses (that can be spoofed or are not fixed by load-balancing). We use two forms of encryption and a custom negotiation protocol to establish a connection, followed by authorization of user (helpdesk) and hardware (Remote Support Client).

Frank

Thanks, that's how I interpreted the diagram you posted. In other words, no need for the 'support server', the same functionality can be built using just a simple black-box running some minimal Linux or BSD distro. Place the box inside the client's LAN, make it accessible from the outside through certificate-authenticated VPN, and give it strict outgoing firewall rules.

The client has to trust that the ruleset that is shown to him is actually the ruleset that the box implements. And as soon as the ruleset includes even one RDP or VNC link to an internal server, all bets are off anyway and anybody with the support password/cert can use that as a stepping stone to other systems.

j.

You are not getting it completely. Still just thinking about VPN's, huh? It's better to think about 3 proxy servers for which you do not know the communication protocol for.

The Remote Support Server is only required when incoming connections at the customer are not allowed. Either because of corporate policy, third party firewalls/routers that cannot be touched or missing coorporation of an IT department. Otherwise the helpdesk could connect directly to the Remote Support Client, which only requires a single TCP port forwarded in the customer's firewall. Nobody knows how to setup a connection to the Remote Support Client. It has a non-standard communication protocol with multiple encryption and negotiation methods. A small program is required to setup a connection, that will act as a proxy server and starts the client software at the helpdesk. User authorization only gives access to the Remote Support Client. Usually devices at the customer have additional security methods, like encryption and authorization.

OK, I get it. Security by Obscurity. Thanks but no thanks.

j.

We gave the source code of the helpdesk program to 3 hackers and they did not manage to establish an authorized connection to the Remote Support Client via the Remote Support Server. So obscure is a much better word than obscurity.