I got to thinking about how it is sometimes possible to use a DMZ setting on one computer in your network and port forward to the other. This question may be "router specific" but I thought it would make for interesting discussion.
About 11 years ago, I set up an old 8-port Equinox router that had a limit of 10 ports it could forward. By the time I had all 4 computers in the office running, all with various services and features that needed to be on each individual workstation, I quickly ran out of ports. I was able to securely configure the web server as a DMZ with some good firewall software (once I ditched IIS for Apache 1.3 for Windows that is). I was able to free up many of the 10 port-forwarding slots that were being used by the web server, so I learned that DMZ and port-forwarding could work together.
Please correct me if I'm wrong... but this is what my understanding of a DMZ setting is:
A DMZ is an Internal IP Address on the network which all inbound traffic defaults to for requests made to the (External) IP Address.
My thoughts on this are, that if you have a machine configured to use Internal IP Address 192.168.0.10 as the DMZ, and port 80 forwarded to the Internal IP Address of 192.168.0.2, is there a way to (or does it happen already) redirect inbound traffic to port 80 from 192.168.0.2 to 192.168.0.10 in the event of a "computer catastrophe" such as hard drive or power supply failure? In other words, is it possible to set up an in-house redundant web server via DMZ and port forwarding? if not, then how else can you provide in-house redundancy from an internal network?
Thanks for the response Phil, but I like details. If the answer is: No, I can't use a port forwarded IP Address and a different DMZ IP Address in case of crashes/hardware failure. Then... what other ways are available?
Maybe it's just my opinion, but if you consider that 364/365 = 99.726% one day a year (24 hours collectively) of downtime in a year, you will only need redundancy for about a week or two throughout the entire year. All I really NEED is a means to keep the server active while you perform reboots, DNS updates, server transfers, stuff like that.
My remote shared server that I pay about $12/month for has a guaranteed up-time of 99.9%, my biggest issue is that even though the server is technically "up", the peak load times cause rather slow performance. Could I create a DDNS record with a very short TTL (20 seconds or less) that would allow for different DNS server destination addressed servers to pick up where others are slacking?
And as always, I usually assume that the budget for a project is practically zero. Even at my own office network, the up-time for an in-house server can realistically be easily 95%. I would think having a backup DNS at home or at a different office would be able to take you up to 99.99% up-time for little to no "additional costs".