On 9 Jul 2002, Cam Turner wrote:

> On Mon, 2002-07-08 at 12:56, Mario Behring wrote:
> > Hi all,
> >
> > The purpose is to transfer data between sites securely using emails, HTTP,
> > FTP, IP (LAN/WAN), etc.
> >
> > Which algorithm should I use: 3DES or SHA-1 ?
>
> Umm. I don't think they compare... 3DES is a Cipher (Stream Cipher? I
> don't have my books here) and SHA-1 is a cryptographic one-way hashing
> algorithm similar to MD5 (but more secure).
>

block cipher.

> So I'd say from your purpose you should use 3DES to ENCRYPT things (so
> that noone can see the contents) or SHA-1 to hash the message before and
> after it is sent to make sure it didn't get modified in transit.
>
> 3DES is vulnerable to certain kinds of attacks.. I seem to recall that
> it's a variant of the meet-in-the-middle attack  (see Applied
> Cryptography by Menezes et al (which he has for free in PDF on his
> website))... So maybe you should consider some other encryption
> algorithm...
>

3DES is pretty solid for the moment,the MITM attack you are talking about
doesn't really make sense for a cipher I beleive you are getting confused
with the problem that some protocols that use 3DES (SSL and such). While
these do suffer from a pretty complicate MITM attack this is not a
reflection on the security of 3DES at all. DES itself has been broken but
that was only due to it's short key length and 3DEs does not suffer from
this problem. There is also the fact the DES has been around for years and
has been beaten on by some of the best crypto people in the word and has
come out of it looking pretty good so you know you have a solid cipher.

If you are looking for something different may I suggest rijndael which is
was choosen as the new AES (Advanced encryption standard) which meant it
had to go under the microscope of the best crypto people in the world.
It offers an advantage over 3des in that you can make the key length
longer if you want to be parnoid. however the one caveat is that while
it has been tested by crypto people it has yet to stand the test of time
out in the wild like DES has. I trust the bruce schneiers of the world to
analyse a cipher for problems but I trust the test of time more.

One thing I do beleive you are not really getting at here is that the
cipher it not going to be the problem in the security here you can always
find a good cipher but the protocol to trasnfer keys, authentication and
authorization issues are going to be what bite you

If someone wantd to get a look at your data and you
are going to use a good cipher they are not going to attack something like
3DES that would be a waste of their time they are going to attack the
protocol, the cipher isn't going to be much good if someone can interecept
the key you use to communicate. I would suggest that you don't try and
design your own one, there are loads of products both free and
commercecial that will do this for you I would look at them first to see
if they fill your needs.

Look for the whole soln not just a little part of it.

>

-- 
----Rory

Reply via email to