Black Box Version 1.0 Network Card User Manual


 
If there is a necessity to divide traffic on the basis of TCP ports, NAT makes it possible to map
local addresses with one external address using TCP load distribution function.
NAT functioning
The NAT technology defines, as it is stated in the RFC 1631 standard, the ways of IP address
translation, used in one network into another network addresses.
There exist three basic principles of address translation: static, dynamic and masquerading.
Static Network Address Translation
With the help of this concept, NAT can organize translation between the same class networks.
(For example, when each of two networks contain one address (mask – 255.255.255.255). This
strategy is the simplest, because the translation can be described by a couple of simple logical
transformations.
Let us cite an example of address translation from two Class C networks – 194.24.90 and
195.60.3. While passing through NAT to the sender’s address field, the packet, addressed from
the host 194.24.90.13 will contain a change in the IP header from 194.24.90.13 to 195.60.3.13.
Dynamic Address Translation
Dynamic translation is necessary when the number of addresses (internal and external) being
translated is different, however, dynamic translation is sometimes used when static translation
does not work. The number of intercommunicating hosts will be limited, in any case, by the
number of free (available) addresses on the NAT interface.
Dynamic NAT is more complicated, because it requires to keep track of intercommunicating hosts
and possibly even of connections, in case when the information (content) must be modified at
Layer 4 (TCP, for example).
For example it is necessary to translate dynamically all IP addresses in Class B network 138.201
into addresses of Class C network 190.200.112. Then, each new connection receives an address
from Class C network if there are available addresses there.
This technology, in contrast with static translation, introduces a new notion – NAT table. It is a
rendition table of internal addresses and NAT-interface addresses (hereinafter, NAT addresses)
Masquerading (NAPT, PAT)
The Port Address Translation is another case of dynamic translation. Here, we have only one
external address behind which, internal addresses “are hidden” – there can be as many internal
addresses as possible. In contrast to the original dynamic translation, PAT does not mean that
there can be only one connection at a time. To multiplex the number of connections, TCP port
information is used by this masquerading. Thus, only the number of ports available limits number
of simultaneous connections.
Version: 1.0 Page. 29 of 95