Isn't this a question of whether you want to route or NAT?

A server that is Internet-facing has to have (or be reachable via) a
public IP. If your ISP changes your block of public IP addresses, you
have to change:

1) the mapping between your private IP addresses and the new public
IP addresses (the static or 1:1 NAT case) or
2) the IP addresses of all the servers, the IPs of the trusted and
external interfaces on the firewall, and the routing table (or
routing protocol configuration)

(2) seems like a whole lot more work to me.

On 21 Jul 2005 at 18:28, David Lang wrote:

> On Thu, 21 Jul 2005, Paul D. Robertson wrote:
> > On Fri, 15 Jul 2005, Matt Bazan wrote:
> >
> >> Is there a preferred method of setting up a Internet facing
> >> screened subnet and the use of public or private IP addresses?
> >> Looking at redesinging our DMZ to only include public resources
> >> (www, smtp, imap, ftp). Presently we use a private IP address
> >> range for this that is NAT'ed at our firewall. Any reasons to
> >> change this policy to using public IPs in the DMZ? Thanks,

> >
> > If you're NATing to your internal network, then a rework is
> > necessary- public stuff should be on its own (preferably) physical
> > subnet.
> >
> > IP addressing doesn't matter much, since you'll be letting stuff
> > through the most likely exploit vectors anyway.

> The thing I've been eharing for years about why NAT is better is that
> you may change ISP's and end up with a new set of IP addresses which
> are easier to change if you NAT.
> this may be true (I've actually never seen anyone acutally DO this),
> but you are trading one-time headaches (which I personally believe are
> no more severe then all the other changes that you need to make when
> changing things, firewalls, DNS, NAT tables, etc) for ongoing overhead
> (performance on your NAT device, troubleshooting, bugs in the NAT
> implementation, overloading of the NAT tables, etc)
> I would definantly have things that server the Internet use public
> addresses, once you get behind that layer and have devices that only
> talk to internal stuff, then make it all private addresses.
> David Lang
> --
> There are two ways of constructing a software design. One way is to
> make it so simple that there are obviously no deficiencies. And the
> other way is to make it so complicated that there are no obvious
> deficiencies.
> -- C.A.R. Hoare
> _______________________________________________
> firewall-wizards mailing list

firewall-wizards mailing list