Koozali.org: home of the SME Server

Wifi DMZ

Offline arne

  • *****
  • 1,116
  • +0/-4
Re: Wifi DMZ
« Reply #15 on: October 10, 2007, 08:19:06 PM »
Quote
Forgot to say i have an accesspoint(linksys) in my lan only to provide internet to a laptop and a pda and no need to use the lan.
Would be nice to put that in a DMZ for security reasons.

Of cource it can be done, and it would require anly a minor modification, if any at all, but to make a script of a few lines to do that is not that easy, as such a project of making such a small configuration script, might not be to popular.
......

Offline raem

  • *
  • 3,972
  • +4/-0
Re: Wifi DMZ
« Reply #16 on: October 11, 2007, 05:15:31 AM »
But is'nt the SME project an open source project ?

Yes, but you overlook the fact that a huge amount of commercially funded  development has gone into making sme server what it is today.
Further to that there is ongoing commercial funding in the form of sponsorship to develop specific new code/functionality, without which we would be missing many of the good parts that the existing sme server has in it.
Further to that there are voluntary monetary donations, that help fund contribs.org and development.

The code base that you or I use today is there only as a result of much commercial input, and as such there are many commercial decisions that affect what code will be developed.
There are simply not enough contributing developers to carry on this project as a purely free of cost open source concept.
Many small & not so small additions are as a result of direct commercial sponsorship by end user businesses who vitally need certain additional functionality and pay a developer to create the code.
I'm aware that at least one of the developers creates code that his clients need and effectively donates that monetary value in coding time to the sme server project.
I'm also sure that quite a few of the key developers contribute a huge amount of personal time freely to this project, to an extent that is much greater than any of us do.
You and I benefit from all of the above "paid for" & donated effort, as the code is open source.
« Last Edit: October 11, 2007, 07:08:45 AM by RayMitchell »
...

Offline arne

  • *****
  • 1,116
  • +0/-4
Re: Wifi DMZ
« Reply #17 on: October 11, 2007, 01:25:41 PM »
Well I think I understand in some way.

For me it is mostly the fun and the challange about is all that is the motivation, so that might be different.

By the way, I was doing some thinking on how the firewall anf the server issues could be simplified so that actually all development could be done more easy.

There will, as I will see it, be a need of a modularization between the "firewall stuff" and the "server stuff" to avoid "mixed stuff" that is to difficult to work with (If it should be an option to have more "radically" options for the firewlling part of it).

Actually I think a lot could be done within the framwork of the existing admin-panel.

Suggestion about how:

1. All firewall functinality is pulled out of the automated interaction with the server functions. (All the existing firewall configuration tool is removed or disabled.)

2. When first installed there is a fixed static basic firewall configuration script with some rather restrictive basic configuration. There should be no automated routines that will change this by thenself and without user interphere.

3. Then ther is build a completely new user panel in the admin-panel. The only sole purpose is to generate a new firewall configuration script based on user input.

4. The firewall configuration panel could then consist of an very easy setup with a red and a green fields wher you can hook of the ports you want opened for the red zone and for the green zone.

5. Also there could be an easy graphic interface for port forwarding.

6. The user panel could have such an easy design that the user could see imidiately what he have closed and what is open and which ports that are forwarded.

7. Then there could be some hook of for other funcrion like "answer to ping", "activate dos protection", etc.


When or if the firewall module does only have to deal with the firewll configuration then the complexity of "the firewall things" should be reduced to only a fraction of what it is in the existing system.   

It should be no (big) problem to give the user a graphical overview and a full controll all the time and it should also be easy to implement inn diverse netfilter specialities.

This should give a bether and increased user control and a more easy and flexible solution.

Seen from my point of view such a project would contain an easy part and a difficult part.

The easy part is to configure a 2 or 3 port firewall. Thats how I would see it the next to nothing part of it.

Then there is the difficult part: How to make a web page interacting with a perl script in such a way that it will generate a text file. (I have no idea, but I guess it should be more easy to just to only generate a text file than interacting with all kind of server functions..)

I believe that all off it can be easily done if things are just modularized a bit so there is a problem area related to the server functions, as one unit and a problems releted to the firewall area, as one unit, then divided int to subareas 1. The web shell for generating the firewall configuration script. 2. The content of the fireall script itself.

The difficulty today is not the firewalling itself, but the way all kind of problems it tightly integrated into each other. With some modularization espessially for the firewall stuff, I think that more could be done and it could be done a lot more easy.


*********
*********

If not any big protests, I think I will post a text based 3 port dmz solution and a fine grained trafic controll in all trafic directions in the relatively near future. The solution is up and running with the SME 7.2 just now, but I will just do some more testing first. Will possibly also make a 2 ports variant with 4 directions firewalling as well. There is only one way to develop a firewall, I think, and that is to test it out. If protests I will not post it.
« Last Edit: October 11, 2007, 05:48:37 PM by arne »
......

Offline raem

  • *
  • 3,972
  • +4/-0
Re: Wifi DMZ
« Reply #18 on: October 12, 2007, 02:46:53 AM »
Well I think I understand in some way

I think you are not yet understanding.
Developers do not have time or resources to recreate something that is already working well, and would only be a LOT of work for little gain.
When something can currently be done another way with minimal effort, no one is going to spend hundreds of hours developing code, especially when it is not being funded by anyone.
If there is a commercial (ie monetary) incentive to develop a 3 port firewall configuration, then it is far more likely to happen, but I would guess that it would involve thousands of dollars of effort (to be compatible with the current design).
I think your suggestions of "five minutes here and there" and "fairly easy to do" are gross understatements.

When time/resources are minimal, then the only approach that is practically viable is smaller incremental steps. Small amounts of effort can tackle a big problem little pieces at a time.


Quote
...I think I will post a text based 3 port dmz solution and a fine grained trafic controll in all trafic directions in the relatively near future.

By all means submit your code, that's what has been asked of you many times.
I would suggest that you explore the viability of developers wanting to use this code before you do further development work.
Post the existing code you have now and then discuss improvements and the best direction to take, in the bug tracker.

...

Offline arne

  • *****
  • 1,116
  • +0/-4
Re: Wifi DMZ
« Reply #19 on: October 12, 2007, 06:33:36 AM »
Well this was only one days work and I thought I should complete it a bit more ..

http://forums.contribs.org/index.php?topic=38812.msg176449#msg176449


(I havent set up all the ports etc right because it was not time for it, but the structure should be there.)
......