>Curtis Call <[EMAIL PROTECTED]> wrote,
>Well NAT does cause problems for some applications for which you will need
>real addresses. I can't think of any business applications off the top of
>my head but being a gamer I've run into this problem when trying to
>host games.
As you very correctly observe, multiplayer games very often break
with NAT. The underlying reason tends to be that these games have
various home-brewed multicast routing protocols that depend on IP
addresses to manage the leaves of the multicast tree.
NAT is part of an even broader problem about "Internet transparency."
Fundamentally, the Internet was designed in accordance with the "end
to end assumption," in which it could be assumed that an IP address
was constant from endpoint to endpoint.
Other things, such as tunneling, encryption, etc., also create this
problem. There are some excellent papers by Brian Carpenter and
Eliot Lear, among others, about the broad problem. Unfortunately, I
can't remember if these stayed at Internet Draft or went to RFC.
There have been discussions at the Internet Activities Board level on
these problems.
Lots of material at the IETF NAT Working Group:
http://www2.ietf.org/html.charters/nat-charter.html, including drafts:
---Traditional IP Network Address Translator (Traditional NAT)
---Protocol Complications with the IP Network Address Translator (NAT)
---NAT Friendly Application Design Guidelines
Some of the protocols that often break are things that have IP
addresses inside application layer packets (SNMP, FTP, DNS),
applications that do redirection (HTTP, FTP, RPC), applications that
do reverse DNS lookup, etc.
In my new book, WAN Survival Handbook, I go through at least 12 kinds
of NAT. Basic NAT, which deals simply with IP packets and TCP/UDP
checksums, is inadequate for lots of applications and/or operational
support of those applications. Realistic "NAT" tends to need upper
layer awareness.
>
>At 08:08 AM 1/10/01 -0500, you wrote:
>
>> I have a question here? Why would anyone use register addresses on their
>>private network, while you can use UN-register addresses like 10.X.X.X ? Do
>>you really need to burn register addresses on a private network?
>>
>>I would like to hear anyone opinion on this subject
>>
>>Brian
>>
>>_________________________________
>>FAQ, list archives, and subscription info:
>>http://www.groupstudy.com/list/cisco.html
>>Report misconduct and Nondisclosure violations to [EMAIL PROTECTED]
>
>_________________________________
>FAQ, list archives, and subscription info:
>http://www.groupstudy.com/list/cisco.html
>Report misconduct and Nondisclosure violations to [EMAIL PROTECTED]
_________________________________
FAQ, list archives, and subscription info: http://www.groupstudy.com/list/cisco.html
Report misconduct and Nondisclosure violations to [EMAIL PROTECTED]
