A Practical Look at Network Addresstranslation
A Practical Look at Network Addresstranslation
Address Translation
A Nokia Horizon Manager White Paper
Notwithstanding any other license agreement that may pertain to, or accompany the delivery of, this computer software,
the rights of the United States Government regarding its use, reproduction, and disclosure are as set forth in the
Commercial Computer Software-Restricted Rights clause at FAR 52.227-19.
TRADEMARKS
Nokia is a registered trademark of Nokia Corporation. Other products mentioned in this document are trademarks or
registered trademarks of their respective holders.
030114
Telephone 1-888-477-4566 or
1-650-625-2000
Fax 1-650-691-2170
Europe, Nokia House, Summit Avenue Tel: UK: +44 161 601 8908
Middle East, Southwood, Farnborough Tel: France: +33 170 708 166
and Africa Hampshire GU14 ONG UK email: ipsecurity.emea@nokia.com
Email: tac.support@nokia.com
Americas Europe
Asia-Pacific
Voice: +65-67232999
Fax: +65-67232897
031014
Static NAT
Static NAT (also called inbound mapping) maps a single private network address, which is
typically the address of a network server, to a single public network address. Static NAT allows
hosts outside of the private network to use a public IP address to access hosts on a private
network. Static NAT is a potential security risk. If the network security policy is configured
incorrectly, the private network device mapped to the public IP address might be fully exposed
to the public network.
Figure 1 illustrates several network servers located on a private network space that have public
addresses associated with them through static NAT at the border firewall.
Private Network
Public Network
192.168.41.17 217.83.3.116
192.168.41.7 217.83.3.117
192.168.41.23 217.83.3.118
192.168.41.11 217.83.3.119
192.168.41.29 217.83.3.120
192.168.41.13 217.83.3.121
192.168.41.31 217.83.3.122
All address translations take place within the policy enforcement point, and the translation
process is transparent to both internal and external entities. When hosts from the public network
try to contact private network hosts, the packets are either dropped or forwarded to the internal
hosts, depending on the security policy.
For an outgoing network transaction, such as in Figure 2, the only change to the packet is the
source address, which is translated to the public network address for the shared network device.
The destination address, source port, and destination port are not modified.
Figure 2 Static NAT for Outgoing Packets
The enforcement point uses a proxy address resolution protocol (ARP) technique to respond to
the request for the addresses that use static NAT. One network device, such as a firewall,
responds to ARP requests intended for another network device. By impersonating the identity of
the targeted network device, the firewall accepts responsibility for routing packets to the true
destination.
Hide NAT
Hide NAT is a type of dynamic NAT that uses different network source ports to map multiple
private addresses to a single public address. This type of address mapping is also known as:
! IP masquerading
! Port address translation
! Single address NAT
! Port-level multiplexed NAT
Regardless of the name, in this type of address mapping, the mapping is not static. In hide NAT,
for each session between an internal network device and the public network, the public IP
address remains the same, but the source port for each device changes. Figure 4 provides an
overview of a network topology that uses hide NAT. The figure also shows an example of the
address translation table.
192.168.67.200
192.168.67.46
Public Network
217.83.3.1
192.168.67.154
192.168.67.73
With hide NAT, address translations do not exist in the NAT table of the enforcement point until
the enforcement point receives traffic that requires translation. Dynamic translations also have a
timeout period. When the period ends, the translations are purged from the translation table,
which makes them available for other internal addresses. The timeout period is different for each
transport protocol (TCP or UDP) and NAT device.
For a network connection, the Check Point VPN-1/FireWall-1 application changes the source
TCP or UDP port of the packet so that it can track the client server traffic throughout the life of
the connection more effectively, as shown in Figure 5.
NAT As a Proxy
NAT operates at the network layer (layer three) of the OSI network model, which is shown in
Figure 6. Since NAT operates below layer four of the OSI model, it is a fundamental proxy. A
fundamental proxy has no cache, and its only function is to take a network connection and
transparently pass it from one side of the network to another.
Figure 6 OSI Reference model
Application interface
Interface to the Host services
User Accounting and Security HTTP, FTP, SSH, Telnet, etc.
Compression/Decompression, Encryption/Decryption
Context/Syntax Data Reformating Data abstraction
Multiplexing, Initiation, Management, and
Application activity Termination of Sessions Communications duplex
Data Handling, End-to-End Reliability TCP, UDP, & ICMP
A proxy is any device that acts on behalf of another device. For example, a Web proxy acts on
behalf of a Web server. Network clients make a requests to the Web proxy, and the Web proxy
makes requests to the appropriate Web server on behalf of the network client. The advantage of
having a proxy is that the Web proxy can cache commonly accessed sites locally. Local caching
reduces outbound traffic, which increases the performance to the Web user.
Unlike a NAT agent, a Web proxy is not a fundamental proxy because it operates at layer four,
the transport layer. A Web proxy is more complex than the type of proxy involved with NAT.
The Web proxy is not transparent and must be explicitly supported by the clients it represents.
The network administrator must configure each client to use the correct proxy. Whenever the
configuration changes, the network administrator must reconfigure every client. Also, the
proxies that operate at layer four and above are typically slower than fundamental proxies.
Table 3 lists the network protocols that the Nokia Horizon Manager v1.3 server uses to
communicate with the managed network devices.
Table 3 Nokia Horizon Manager v1.3 Server and Managed Network Device Protocols
Port name Port number Description
Secure shell 22/tcp Secured managed device communication (ssh and scp)
Nokia Horizon Manager is an application that manages network policy enforcement points. All
of the network devices that Horizon Manager manages should have routable, or at least
reachable, IP addresses. Moreover, the network protocols listed in Table 3 are common network
services and are typically allowed on the private network. The network protocols are required for
Horizon Manager to manage the network devices.
You must add the four network protocols listed in Table 2 to any security policy at intervening
policy enforcement points. The RMI_Server_port protocol is a bidirectional communication
mechanism, while the RMI_Registry_port and SSL_Server_port network protocols are mono-
directional (client to server only). This means that the client or the server can both initiate
network communication for the Horizon Manager application.
Figure 7 illustrates several potential Horizon Manager client locations with respect to the
Horizon Manager server and NAT gateways. The basic configuration has two Horizon Manager
clients within the central enterprise network. One Horizon Manager client is on the same subnet
as the Horizon Manager server. As long as the enterprise name server can resolve the host names
for the Horizon Manager client and the Horizon Manager server, the client-server pair can
communicate.
In Figure 7, there is also a Horizon Manager client behind an internal router in the central
enterprise network. Since the internal router does not translate addresses or block any network
traffic, the Horizon Manager client and the Horizon Manager server can communicate. Again,
the enterprise name server must be able to resolve the host names for the Horizon Manager client
and the Horizon Manager server to authenticate each other.
Figure 7 Horizon Manager Client Connections to the Server Through Different NAT
Conditions
All the network device addresses belonging to this subnet are translated
at the firewall. Since, the majority of the network devices in this subnet
are clients and have their addresses dynamically assigned, this is the
best solution. The NHM Client's private network address is translated
using Static NAT to make it accessible from the enterprise network.
VPN Tunnel
de
vi
NHM Client NHM ce NHM Client
Server m
An IPSec VPN tunnel provides the only an
ag
secure mechanism with which to access em
the enterprise's NHM server through the en
t
public network.
NHM ClientCentral
This NHM client is on the
same subnet as the Enterprise
Network
enterprise's NHM server.
Another Horizon Manager client in Figure 7 is in an isolated subnet outside of the central
enterprise network, but within the enterprise. This subnet contains dynamically assigned RFC
1918 addresses for all the desktop workstations within the subnet. To prevent address collisions
with any other part of the enterprise network, the firewall in the subnet uses hide NAT to mask
all addresses behind the network interface of the firewall that faces the enterprise.
Hide NAT at the subnet firewall prevents the Horizon Manager client from communicating with
the Horizon Manager server. More importantly, hide NAT prevents the Horizon Manager server
from connecting to the Horizon Manager client (RMI_Server_port, 1201/tcp). To allow
communication between the client and server, you must configure the firewall to statically
translate the address of the Horizon Manager client in the isolated subnet.
The Horizon Manager client located outside of the enterprise network requires a VPN tunnel to
securely communicate with the enterprise Horizon Manager server. To remain secure, the
enterprise Horizon Manager server should never be exposed to the public network. Even though
all communication between the Horizon Manager server and any Horizon Manager client is
encrypted (128-bit SSL), if hosts on the public network can see the Horizon Manager server, the
server becomes a potential target for attack.
If the gateway at the remote site uses hide NAT for the hosts behind it (for example, to maximize
network resources), the gateway might drop the Horizon Manager communication protocols.
Any gateway between the Horizon Manager client and Horizon Manager server that uses hide
NAT blocks Horizon Manager server-to-client communication. When the Horizon Manager
client and server communicate through an IPSec VPN tunnel rather than over the public
network, the communication is secure, and the traffic between the client and the server is not
dropped by any gateways.
Summary
One of the most important reasons to use NAT is that it simplifies network administration. NAT
gives you flexibility in assigning internal network IP addresses since it allows you to separate
public network spaces from private network spaces. For example, if you move a Web server or
FTP server to another host, you only need to change the address mapping at the firewall to
reflect the new host. Also, if you make changes to the internal network, you do not need to
reconfigure the IP address for each host because the public IP address for each device in the
internal network either belongs to the firewall or is associated with external network interface of
the firewall.
NAT is also cost-effective. The non-public routable address blocks defined in RFC 1918 can be
reused repeatedly without an associated registration cost. With hide NAT you can use a single
network device to provide network access for nodes within a private network space.
When you use NAT, make sure you do not block client-server communication for applications
like Horizon Manager v1.3.
While NAT is practical, you should never use it as a method to secure a private network. NAT is
only capable of hiding network devices, not securing them. The first step in securing the network
is to put a firewall at the border of the network.