Re: Pseudorandom Flow Labels

[Fernando Gont <[e-mail]>]

Thomas,

On 05/04/2011 05:36 p.m., Thomas Narten wrote:
> Case in point about how we are being *extremely* loose in using the
> term "pseudo random".
[....]
> Part of my objection to the term "pseudo random" is that the term has
> not been defined within the context of the Flow Label.

You raise a very good point, indeed. For instance, when we talk about
e.g. "port randomization", we're really talking about "producing port
numbers that are unpredictable by off-path attackers".

To make this terminology issue worse, it has been argued a few times (by
some mathematician IETFers) that the properties that we need for the
"hash" functions in the hash-based algorithms are really that of PRFs
(Pseudo Random Functions) (i.e., hash functions being a specific example).

In summary, I agree with the terminology issue that you've raised. I'd
probably argue that the best way to go is to specify which properties we
want for Flow Labels, such as they have been specified for port numbers
in RFC 6056. Namely:

* We want Flow Labels that unpredictable by off-path attackers (history
has taught us that this is a good proactive measure)
* We want an algorithm for generating FL that produces FLs that do not
repeat with a high frequency (i.e., they are distributed normally)

One possible algorithm for achieving these properties is calling a
random()-like function. But there are others, such as the hash-based
algorithms specified in draft-gont-6man-flowlabel-security.

Thanks,
-- 
Fernando Gont
e-mail: [e-mail] || [e-mail]
PGP Fingerprint: 7809 84F5 322E 45C7 F1C9 3945 96EE A9EF D076 FFF1