There's this idea of using 2 brands of firewalls both inline that is usually brought up:

  {Internet}<-->FW(A)<-->FW(B)<-->{Internal Network}

Or using 2 algorithms:

  {Plain Text}<-->Alg1(e.g AES)<-->Alg2(Custom Algorithm)<-->{Cipher Text}

And this is usually done in a way that one of the algorithms or hardwares used are public and known brands and the other one is a local brand or custom made device/algorithm. For example they use AES and Alg_x where Alg_x is not a publicly known algorithm and is made by government of country X.

The rationale behind this is that if country Y which is at war with country X puts a backdoor in the publicly known device/algorithm the second device/algorithm made exclusively by government "X" will keep the data protected and if there are some missing features in that, the first device will complement it.

Or in another case I use an American brand device along with a French/Russian/Chinese/... Device. So if there's backdoor in both of these devices, those 2 governments are most likely not willing to share their access.

  • Is this a common practice? Has any organization or standardization institute referred to this? (Protecting your vital resources with at least 2 types/models/brands of security devices.)

Edit: slightly changed the questions to make it fit our Q&A.

  • 3
    "I have a firewall made by myself which [...] I'm sure is free of backdoors" -- really? I think it's highly unlikely that you can guarantee that. Did you make the BIOS/hardware/firmware yourself? Did you write the source for the kernel yourself? Did you write the source for every service running on your machine yourself?
    – Chris Down
    Feb 21 '16 at 12:06
  • @ChrisDown Remember I'm talking from the perspective of a military organization, they often develop the firmware themselves, they use specific hardware either made by trusted parties or undergone strict tests that ensure them to a good degree that they are free of backdoors.
    – Silverfox
    Feb 21 '16 at 12:09
  • 2
    I think this question is too open and will leave too much space for opinion Feb 21 '16 at 12:21
  • @ChrisDown Theres also the possibility that 2 different parties have access to each of the backdoors and they are not willing to share information to gain access. (and each device is blocking the other devices backdoor)
    – Silverfox
    Feb 21 '16 at 13:50

Funnily enough I have seen this technique recommended in some NSA best practice documentation pre-Snowden. See #1 under Network Recommendations in Best Practices for Keeping Your Home Network Secure. It refers to routers and not firewalls, but most routers are also operating as firewalls.

I believe this is mostly academic though as I would like to point out that while yes deliberate back-doors in both products will be difficult to exploit without knowledge of both back-doors, you're neglecting the more simple occurrence that it's possible there's vulnerabilities in both, some perhaps even yet to be discovered that are just down to sheer coding error like buffer overflows or as complicated as the shellshock vulnerabilities. With enough expertise and time I'm sure some vulnerabilities could be discovered in either product, back-door or no back-door.

A lot of routers/firewalls will end up using Linux and likely some version of OpenSSL. So any vulnerabilities discovered in both will still likely be exploitable in both devices.


This assumes both implementations do not have similar errors. Libc, openssl etc. Would longer chains work better and if so by how much?

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.