Re: DBMS security assisted with crypto h/w?

Thierry,
You think that trusting cryptography to the machine is a solution for all e-commerce problems, don't you ?
I am sorry, but you're wrong.
1) Can you be completely sure that those hardware devices do not  have a 'secret' backdoor, that has been left opened accidentally,  intentionally, or was insisted by authorities ? And do you think  that in case this backdoor exists hackers will never be able to find it ? 1.5) When you have an open-source crypto package, you can always (probably after the attack) verify the code and find a possible hole. But I'd like to watch you verifying the hardware chip. :-) 2) transaction authentication based on closed-key approach will work only if you have a very limited number of correspondents (ideally 1) and have the same strength as simple password identification.
3) Imagine you have an ideal authentication procedure and an ideal channel. What information would you get on your side ? You will precisely know that there is a right _key_ there. But it is not key who works, earns money and spends them. This is the biggest problem with e-commerce today - instead of verifying buyer's identity and validating his money, seller verifies authencity of the buyer's wallet (!)
> Am I dreaming of a level of e-commerce security which is beyond market
> expectations?

You are dreaming of a level of e-commerce security which is far below market expectations.The whole problem is much more complicated than just cryptography solution.

With best wishes
Alexander

Thierry Moreau <Thierry.Moreau_at_connotech.com> wrote in article <37D019C6.4C49_at_connotech.com>...
> Hi,
>
> I'm looking for DBMS security assisted with cryptographic hardware.
>
> More precisely, suppose an application uses secret cryptographic keys
> that are used to authenticate transactions from remote systems (e.g.
> with a triple-DES MAC computation, no public key cryptography involved).
> I would expect DBMS supplier to team up with cryptographic hardware
> suppliers to provide a seamless solution so that the remote systems keys
> are stored in encrypted form in the DBMS, and each and every
> cryptographic computation is made in the cryptographic hardware. Then no
> application programmer can use a remote system key on the server, and
> the service provider can claim that "insider fraud" is a *very* remote
> possibility.
>
Received on Mon Sep 06 1999 - 02:56:20 CDT