Security Basics mailing list archives
Re: DMZ Design and Functionality
From: Schneider Sebastian <ses () straightliners de>
Date: Thu, 21 Aug 2003 01:46:25 +0200
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 Hi all folks ;-) I guess I didn't get your set up quite right. Actually a DMZ is placed between the Internet and your local network. So I assume you're all talking about screened subnets. There are different designs depending on the traffic volume you currently observe or that will occure respectively. In general, if you're quite familiar with RedHat you should stick to that one for your two new hosts. Second, when running a firewall you should know which one you're about to implement, since there are several types: packet filters, stateful firewalls and proxy firewalls. Packet filtering/stateful firewalls can be implemented using the open source iptables/netfilter ( http://iptables.org/ ). Proxy firewalls can be implemented transparently to the user and provide the advantage of inspecting the payload of packets sent thru/to your network. In the latter case, you have to check your Security Policy and regional/national laws regarding data protection/privacy. It's kind of hard to measure the hardware requirements for they depend on different criteria as inbound/outbound traffic, mails processed, broadcasts, anti-virus / firewalling software running. You should also take account of planning the requirements regarding future increase expected. Placing a spam filter like SpamAssasin and a virus mail filter on the same machine usually is no problem, depending on the machine running these. A client of us experiences about 7,000 mails inbound a day AV/UCE-filtered by one host based on a P4 1.8 GHz processor (IBM xSeries 330). Even at peaks, there is no noticeable delay processing the messages. Why is it, you're going to place the hardware firewall second tier? Commonly they are tuned to process large amounts of traffic and applying the rules set up. You might use it to diburden your main firewall. Are there any other hosts which have to be made availabe to the public? Do they experience high traffic? There are three ways to design your network. The first is applying one firewall behind every Internet/WAN connection. One NIC is handling traffic from/to the remote network, one NIC to your screened network and a third to your local network. The second is inlining firewalls, one behind another. That is placing one firewall behind your Internet/WAN connection, then your public servers, second firewall and after all your local network. This set up has the advantage of applying different access control restrictions to different security zones/subnets - the closer the subnet to the Internet the less the security level. Setups like that are not that odd as it sounds but hard to troubleshoot, however. The third design is placing two firewalls equidistant from the the Internet, One application gateway to your public services and one firewall to your corporate network. Those considerations depend on your current design, business needs and services you want to offer as well as your Security Policy. I guess you prefer the first one maintaining a local network and a screened subnet. Usually screened subnets don't really power down the overall performance. You should place your mail filter (AV/UCE) into the screened subnet working as a mail relay (as known as front-end mailer). Inbound messages will get forwarded to your internal mail server and outbound to your mail filter in the screened subnet. Dropped messages - recognized as spam or virus-containing - should be stored to a "special" account, since some users might scream up because of mails missing. VPNs are kind of hard to choose, since there are several solutions. If you prefer IPSec based VPN connections, please notice not all protocol options are capable traversing NAT gates. IPSec with ESP in tunnel mode should work out as well as AH in transport mode. There are different points to keep track of. Are there a lot of teleworkers / mobile users / business partners? Which key configuration are you up to? Is the device decrypting the enciphered traffic placed behind or before your firewall? What is stated in your SecPol? Pre-shared keys are easily to maintain, since everyone is using the same, but it's like giving them all a password. Using certificates is easy in administration, since digital certificates are assigned seperately on a connection-by-connection basis. But a bit more complicated to set up, since PKI is commonly used in those environments. More about IPSec can be found at http://www.freeswan.org/ , http://www.rommel.stw.uni-erlangen.de/~hshoexer/ipsec-howto/HOWTO.html , http://www.freeswan.ca/docs/freeswan-1.5/doc/links.ipsec.html#opensource . As Chris pointed out, the border firewall as well as VPN endpoint can be implemented using a small hardware appliance like a Cisco device. This depends as well if you're poviding RAS services. Building up VPNs with Cisco is pretty cool and (almost) kind of easy. You should also make up your mind, if IDS sensors are going to be placed on your network. That has also an impact on how to design what. If you have further questions, feel free to e-mail or giving a buzz. - -- straightLiners IT Consulting & Services Sebastian Schneider Metzer Str. 12 13595 Berlin Germany Phone: +49-30-3510-6168 Fax: +49-30-3510-6169 Diese E-Mail enthält vertrauliche und/oder rechtlich geschützte Informationen. Wenn Sie nicht der richtige Adressat sind oder diese E-Mail irrtümlich erhalten haben, informieren Sie bitte sofort den Absender und vernichten Sie diese Mail. Das unerlaubte Kopieren sowie die unbefugte Weitergabe dieser Mail ist nicht gestattet. This E-Mail may contain confidential and/or privileged information. If you are not the intended recipient (or have received this E-Mail in error please notify the sender immediately and destroy this E-Mail. Any unauthorized copying, disclosure or distribution of the material in this E-Mail is strictly forbidden. On Tuesday 19 August 2003 09:21, Meidinger Chris wrote:
Hi Dana, i agree with David that it's a pretty advanced approach, but assuming you have no time pressure, it's a sound infrastructure. Just be sure you don't promise anyone when it will be in production. One thing i would change in your place is i would put both the firewall and the proxy/mail on the same operating system. It will be enough administration if you have two new *nix boxes (assuming your background is not unix) without you having to keep up on patches/updates/administration for two operating systems. How you size the firewall machine depends on the width of your internet connection. What kind of a connection do you have? Now, if your firewall is going to be fairly simple, you might even want to look into an inexpensive hardware firewall. Assuming it would cost you say 2500$ (maybe a low estimate) for a BSD machine that you would have to administer constantly, you could already get a (smallish) hardware firewall for that money. Sizing the proxy/mail machine will also depend on your web/mail traffic. We have no idea how big your site is/what connextion you have/how much mail traffic you have. If you want to do a serious VPN solution, then a hardware firewall instead of the BSD machine makes even more sense. If you can get your company to spring for it, get a CheckPoint FireWall-1 on Nokia with VPN-1. These things can all be done in software on a self-installed OS, but if you are alone setting everything up, a hardware solution will be to your advantage in terms of time and manageability. I hope i answered all your questions. If i was wrong on any point, then list, please let me know. badenIT GmbH System Support Chris Meidinger Tullastrasse 70 79108 Freiburg -----Original Message----- From: Dana Rawson [mailto:absolutezero273c () nzoomail com] Sent: Monday, August 18, 2003 9:53 PM To: security-basics () securityfocus com Subject: DMZ Design and Functionality Forgive me if these questions are too basic but I am relatively new to this. I am the network administrator at my company and over the past year have become aware of a need for increased security. I have been reading posts here in hopes of learning more about this. While I have learned considerable amounts, and have searched for answers elsewhere, I am still in need of guidance. Any help or direction would be greatly appreciated. I am open to reading any books that one might recommend. I have seen a few books out there but not sure which are worthwhile. Anyway, my background information is this: I wanted to install a DMZ at 2 of my company's locations. I do have a limited budget so I was planning on using OpenBSD for my first tier firewall. I do have a hardware based firewall in place currently which I was planning on using as my second tier firewall. My initial plan is to build a machine using OpenBSD that does nothing but firewall. Additionally, I wanted to add another machine to run Sendmail/SpamAssassin and an an anti-virus software. On this I was hoping to run Redhat as this is what I am most knowledgeable on. My thought behind this was to block spam, of course, and also run a gateway anti- virus solution that would block viruses coming from websites and employee's personal e-mail accounts. This due to the fact that I have seen a number of viruses coming in from either their 'webmail' or through their Outlook Express. I wish to set up an ftp server and webserver to facilitate OWA. Additionally I would like to make available VPNs and encrypt all data transmitted over remote connections. Remote connections may consist of a MS RAS and Citrix. With this information my questions are: 1. To begin, does this sound like an acceptable solution? 2. How do I size the machine that I am going to run OpenBSD? I have read that a DMZ will slow performance down some. If I have a fast enough machine will it aid performance? At what point is overkill when running OpenBSD. 3. How do I size the machine that will be running Redhat, Sendmail and SpamAssassin? Is this configuration acceptable? Should the Anti-virus software be running on a separate machine? 4. What open source options to I have for encryption and VPNs? 5. Are there any potential problems running this configuration? Does everything mentioned here play nice together? Would you change anything here and if so why? Many thanks in advance. Dana --------------------------------------------------------------------------- --------------------------------------------------------------------------- - --------------------------------------------------------------------------- --------------------------------------------------------------------------- -
Diese E-Mail enthält vertrauliche und/oder rechtlich geschützte Informationen. Wenn Sie nicht der richtige Adressat sind oder diese E-Mail irrtümlich erhalten haben, informieren Sie bitte sofort den Absender und vernichten Sie diese Mail. Das unerlaubte Kopieren sowie die unbefugte Weitergabe dieser Mail ist nicht gestattet. This E-Mail may contain confidential and/or privileged information. If you are not the intended recipient (or have received this E-Mail in error please notify the sender immediately and destroy this E-Mail. Any unauthorized copying, disclosure or distribution of the material in this E-Mail is strictly forbidden. -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.2.2 (GNU/Linux) iD8DBQE/RAhSQ7mOWZBxbPcRAgO3AKCqoT6LWpK+CF+Kfo35inS7sp4M6ACglEYG 6eQ353TtUWlaMjwUtIf3Rb4= =oUjQ -----END PGP SIGNATURE----- --------------------------------------------------------------------------- ----------------------------------------------------------------------------
Current thread:
- DMZ Design and Functionality Dana Rawson (Aug 18)
- RE: DMZ Design and Functionality David Gillett (Aug 18)
- <Possible follow-ups>
- RE: DMZ Design and Functionality Meidinger Chris (Aug 19)
- Re: DMZ Design and Functionality Schneider Sebastian (Aug 20)