funsec mailing list archives

Re: Can you trust Chinese computer equipment?


From: "Tomas L. Byrnes" <tomb () byrneit net>
Date: Sun, 14 Feb 2010 16:52:26 -0800



-----Original Message-----
From: Valdis.Kletnieks () vt edu [mailto:Valdis.Kletnieks () vt edu]
Sent: Sunday, February 14, 2010 4:15 PM
To: Tomas L. Byrnes
Cc: Benjamin Brown; funsec () linuxbox org
Subject: Re: [funsec] Can you trust Chinese computer equipment?

On Sat, 13 Feb 2010 21:48:31 PST, "Tomas L. Byrnes" said:
The corollary of the "test baseline" in my prior post is that EVERY
piece of hardware that comes into my networks gets reflashed and
reloaded with MY gold master disks/config.

That just pushes the problem around.  How do you know that basically
unaudited IOS you just flashed into that Cisco doesn't have a very
subtle back door in it, left by some Chinese-agent coder (who could
possibly be a disgruntled white dude) back in IOS 11?

And yes, there are organizations where that level of tinfoil-hat
paranoia is called for...

Not only does this eliminate preinstalled malware, but I also get
zero
crapware going into production.

It's hard enough to find a version of IOS that actually *works* - most
sites end up settling on one that only has non-debilitating issues in
their environment.

But the fact that Cisco box is probably not loaded with the one IOS
version that actually works in your network is reason enough to
reflash
it.

Security is a degenerate case of traffic and configuration
management.

I'd hassle you about that one, except that Verizon study that showed
config issues contributed to 90% of the breaches.  Damn pesky facts.
;)
[Tomas L. Byrnes] 
[Tomas L. Byrnes] 
We're arguing degree here. It's about controlling what you can control
so you have time and energy to focus on what you can't, and baseline and
positive control of what gets installed is one of them.

Agreed, if the scenario Winn Schwartau wrote about in "Terminal
Compromise", where the author of the dominant OS was compromised so that
a foreign power can put a bomb in everyone's machine, then no-one,
except those who have complete source code to object code, and the full
build cycle, control of everything, with every one who has access
subject to least privilege and independent audit at each stage, is safe.

I've worked in places where that sort of tinfoil hat is necessary. It's
why we get "$5,000.00 toilets", since it's much safer to not allow any
exceptions to the audit protocol than to run the risk of something
critical not being subject to it.

For mere mortals, making everything your own "Gold Master" before
deployment is a no duh step, and has the added benefit of preventing
pesky interoperability problems and being "bug compatible", and behaving
in the way your sysadmins are used to.

As for IOS, well, IMO, it continues to exist for the same reason as
Windows, inertia (which isn't an invalid reason, btw).





_______________________________________________
Fun and Misc security discussion for OT posts.
https://linuxbox.org/cgi-bin/mailman/listinfo/funsec
Note: funsec is a public and open mailing list.


Current thread: