Re: Crypt data "on the fly"
On Tue, 3 Jun 2014 10:03:17 +0100
Darac Marjal <firstname.lastname@example.org> wrote:
> I like the idea of Crashplan, but just slapping the label of
> "Blowfish" on their encryption isn't quite good enough .
Mouhaharf, trusting SO for trivial Q/A is one thing, trusting
it about things as sensible as crypto is much more neurone
instant death than common sens…
> If I could trust that the encryption was done competently, I'd
> move to Crashplan in a heartbeat.
Most of the programmers use directly the code supplied by Schneier's
But you're right: when you don't know how the crypto's implemented
don't ever use it.
> Yes, but choosing your encryption badly can cause problems. For
> one, as above, bad choices can mean poor security.
So, you're a real cryptanalyst; then, please develop your
rant about BF.
> But also a
> badly chosen encryption scheme might mean unnecessarily large
> diffs (and so more storage/bandwidth on your cloud provider).
Depends what you're looking after: real security or comfort…
At this time, there's no known successful attack against BF.
thinking 448 bits is 3.5 more secure than 128 bits doesn't
mean nothing (AES 256 bits is considered weaker than 128 bits…)
provided you have a strong random data generator (otherwise
not any crypto will long more than a few hours).
As of today, 128 bits fits all the needs.
Much of people saying: "oh, this cryto's not so good", etc
are either jealous or disinformers, unless they are specialists
and publish academic papers; a very few are real cryptanalyzers
because that needs real maths applied to crypto skills.
This is no secret that nsa pushed AES in front of BF, there
are good reasons for that…
BF have some weakness (some keys aren't that safe) but nobody
has broken the whole set of rounds, and even if it is an "old"
crypto (it works on 8bits µCPU!), it is still one of the best
there is at this time (and a polyvalent one).
You will not censor me through bug terrorism.
-- James Troup