Windows Server 2008 R2 : Deploying and Enforcing a Virtual Private Network (VPN) Using an RRAS Server

4/21/2011 3:45:25 PM

A common method of securing information sent across unsecured networks is to create a virtual private network (VPN), which is effectively a connection between two private nodes or networks that is secured and encrypted to prevent unauthorized snooping of the traffic between the two connections. From the client perspective, a VPN looks and feels just like a normal network connection between different segments on a network—hence the term virtual private network.

Data that is sent across a VPN is encapsulated, or wrapped, in a header that indicates its destination. The information in the packet is then encrypted to secure its contents. The encrypted packets are then sent across the network to the destination server, using what is known as a VPN tunnel.

The Windows Server 2008 R2 RRAS role allows for the creation of VPNs, and integrates with the NPS role to provide for validation of client health before creating a VPN session.

Note

Virtual private network support in Windows Server 2008 R2 provides for simple VPN tunnels to be created. For more complex scenarios where specific rules need to be created and application-layer filtering of the VPN traffic is needed, look at Microsoft’s Forefront Edge line of products, which includes the Forefront Threat Management Gateway (previously called Internet Security and Acceleration or ISA Server) and the Forefront Unified Access Gateway products.

Exploring VPN Tunnels

The connection made by VPN clients across an unsecured network is known as a VPN tunnel. It is named as such because of the way it “tunnels” underneath the regular traffic of the unsecured network.

VPN tunnels are logically established on a point-to-point basis but can be used to connect two private networks into a common network infrastructure. In many cases, for example, a VPN tunnel serves as a virtual wide area network (WAN) link between two physical locations in an organization, all while sending the private information across the Internet. VPN tunnels are also widely used by remote users who log on to the Internet from multiple locations and establish VPN tunnels to a centralized VPN server in the organization’s home office. These reasons make VPN solutions a valuable asset for organizations, and one that can be easily established with the technologies available in Windows Server 2008 R2.

Note

VPN tunnels can either be voluntary or compulsory. In short, voluntary VPN tunnels are created when a client, usually out somewhere on the Internet, asks for a VPN tunnel to be established. Compulsory VPN tunnels are automatically created for clients from specific locations on the unsecured network, and are less common in real-life situations than are voluntary tunnels.

Tunneling Protocols

The tunneling protocol is the specific technology that defines how data is encapsulated, transmitted, and unencapsulated across a VPN connection. Varying implementations of tunneling protocols exist, and correspond with different layers of the Open System Interconnection (OSI) standards-based reference model. The OSI model is composed of seven layers, and VPN tunneling protocols use either Layer 2 or Layer 3 as their unit of exchange. Layer 2, a more fundamental network layer, uses a frame as the unit of exchange, and Layer 3 protocols use a packet as a unit of exchange.

The most common Layer 2 VPN protocols are the Point-to-Point Tunneling Protocol (PPTP) and the Layer 2 Tunneling Protocol (L2TP), both of which are fully supported protocols in Windows Server 2008 R2.

PPTP and L2TP Protocols

Both PPTP and L2TP are based on the well-defined Point-to-Point Protocol (PPP) and are consequently accepted and widely used in VPN implementations. L2TP is the preferred protocol for use with VPNs in Windows Server 2008 R2 because it incorporates the best of PPTP, with a technology known as Layer 2 Forwarding. L2TP allows for the encapsulation of data over multiple network protocols, including IP, and can be used to tunnel over the Internet. The payload, or data to be transmitted, of each L2TP frame can be compressed, as well as encrypted, to save network bandwidth.

Both PPTP and L2TP build on a suite of useful functionality that was introduced in PPP, such as user authentication, data compression and encryption, and token card support. These features, which have all been ported over to the newer implementations, provide for a rich set of VPN functionality.

L2TP/IPSec Secure Protocol

Windows Server 2008 R2 uses an additional layer of encryption and security by utilizing IP Security (IPSec), a Layer 3 encryption protocol, in concert with L2TP in what is known, not surprisingly, as L2TP/IPSec. IPSec allows for the encryption of the L2TP header and trailer information, which is normally sent in clear text. This also has the added advantage of dual-encrypting the payload, adding an additional level of security into the mix.

L2TP/IPSec has some distinct advantages over standard L2TP, namely the following:

L2TP/IPSec allows for data authentication on a packet level, allowing for verification that the payload was not modified in transit, as well as the data confidentiality that is provided by L2TP.
Dual-authentication mechanisms stipulate that both computer-level and user-level authentication must take place with L2TP/IPSec.
L2TP packets intercepted during the initial user-level authentication cannot be copied for use in offline dictionary attacks to determine the L2TP key because IPSec encrypts this procedure.

An L2TP/IPSec packet contains multiple, encrypted header information and the payload itself is deeply nested within the structure. This allows for a great deal of transport-level security on the packet itself.

Enabling VPN Functionality on an RRAS Server

By installing the Routing and Remote Access Service (RRAS) on the server, the ability to allow VPN connections to and/or from the server is enabled. The following type of VPN connections can be created:

VPN gateway for clients— The most common scenario, this involves the RRAS server being the gateway into a network for VPN clients. This scenario requires the server to have two network cards installed.
Site-to-site VPN— In this scenario, the RRAS server creates a VPN tunnel between another RRAS server in a remote site, allowing for traffic to pass unimpeded between the networks, but in an encrypted state.
Dial-up RAS server— In this layout, the server is installed with a modem or pool of modems and provides for dial-in capabilities.
NAT between networks— On an RRAS server installed in Routing mode, this deployment option provides for Network Address Translation (NAT) between network segments. For example, on one network, the IP addresses might be public, such as 12.155.166.x, while on the internal network they might be 10.10.10.x. The NAT capability translates the addresses from public to private and vice versa.
Routing between networks— On an RRAS server installed in Routing mode, this deployment option allows for direct routing of the traffic between network segments.
Basic firewall— The RRAS server can act as a simple Layer 3 router, blocking traffic by port. For more secure scenarios, use of an advanced Layer 7 firewall such as Microsoft’s Forefront Threat Management Gateway (previously called Internet Security and Acceleration or ISA Server) is recommended.

Note

Setting up a VPN connection requires the server to have at least two network cards installed on the system. This is because the VPN connections must be coming from one network and subsequently passed into a second network, such as from the demilitarized zone (DMZ) network into the internal network.

To set up the RRAS server for the most common scenario, VPN gateway, perform the following tasks:

1.	Open the Routing and Remote Access MMC tool (Start, All Programs, Administrative Tools, Routing and Remote Access).
2.	Select the local server name or connect to a remote RRAS server by right-clicking Routing and Remote Access and selecting Add Server.
3.	Click Action, Configure and Enable Routing and Remote Access.
4.	Click Next at the Welcome page.
5.	Choose from the list of configuration settings, as shown in Figure 1. Different scenarios would require different settings. For example, if setting up a site-to-site VPN, you should select the Secure Connection Between Two Private Networks option. In this case, we are setting up a simple VPN, so we select Remote Access (Dial-up or VPN). Figure 1. Enabling VPN functionality.
6.	On the Remote Access page, check the box next to VPN. If enabling dial-up, such as in scenarios when the VPN box has a modem attached to it, the Dial-up box can be checked as well. Click Next to continue.
7.	On the VPN Connection page, shown in Figure 2, select which network card is connected to the network where VPN clients will be coming from. This might be the Internet, or it might be a secured perimeter network such as a DMZ. Click Next to continue. Figure 2. Specifying the network card for VPN clients.
8.	On the IP Address Assignment page, select how VPN clients will get their IP addresses (typically Automatically). In addition, a manual range can be specified. Click Next to continue.
9.	On the Managing Multiple Remote Access Servers page, shown in Figure 3, select whether to use RRAS to authenticate locally or to use a remote RADIUS server. Click Next to continue. Figure 3. Specifying RADIUS settings for the VPN server.
10.	Review the wizard settings and click Finish when complete.
11.	Click OK when prompted about the default connection request policy being created and click OK again if prompted about the DHCP Relay Agent.
12.	Click Finish when the wizard is complete.

The wizard will enable RRAS on the server and allow for administration of the VPN settings and client from the Routing and Remote Access dialog box, shown in Figure 4. Review the settings within this tool to familiarize yourself with how the system is configured.

Figure 4. Administering the server from the RRAS MMC tool.

Modifying the RRAS Network Policy

After installing and configuring RRAS, the NPS system will deny access by default to the RRAS server for clients, unless the network policy generated is modified. The network policy, which is labeled Connections to Microsoft Routing and Remote Access server, can be found under the Network Policies node of the Network Policy Server.

The policy must be set to Grant Access in the Access Permission section of the dialog box, as shown in Figure 5 . This dialog box can be invoked by right-clicking the policy and choosing Properties. After enabling, the NPS system will allow client connections.