What you are doing is what banks have been doing for ages with wire
transfers and it is called MAC, Message Authentication Code (not related to
an Ethernet MAC address at all).
Wire transfers are sent in clear text amongst banks. Each bank has set a
'signature' (a code) with each other. They use an algorithm which includes
the message itself and that code. Notice that the message is send in clear
text. It doesn't matter that MD5 is not secure or that it can be decripted
(which, in fact, it cannot, since it is a one-way code), the important point
here is that the extra code appended to the clear message is never found.
Actually, amongst banks not only the message is sent in clear text but the
algorith is well know, the only thing that is not known is the validation
code, which is changed every so often so that even if found out, it cannot
be used for long.
Just make sure that you have a safe means of exchanging keys in between the
servers every now and then. Banks usually send a book of keys for a certain
period physically amongst them and only when they are received and it is
certain they have not been intercepted or tampered with they get used.
Satyam
----- Original Message -----
From: "Peter Lauri" <lists@xxxxxxxxxxx>
To: <php-general@xxxxxxxxxxxxx>
Sent: Tuesday, September 05, 2006 11:04 AM
Subject: Is this unsecure?
Hi,
I have bumped into a problem. I need to use a web service that is located
on
server B from server A. The server B will execute a script when the web
service is accessed and an email is sent as an parameter. The problem is,
if
I only have the email as incoming parameter, anyone can just figure out
the
url for the web service, the name, and then just send the email to that
address.
To make this a little bit secure I setup so two parameters are sent, the
email and a confirmation code. First I was just thinking to basically have
a
password sent with, and if that is correct just execute the script.
However,
due to server restrictions I can not run it on HTTPS, so that also looses
value.
So this is how I solved it:
I send a parameter with the request that is the email, some extra
characters
and then MD5 on that. I do this on server A and then server B just checks
if
it is the same resulting string. If so, we know it comes from server A
because that server is the only one that knows the extra characters used.
$authstring = md5("asdf".$email."fdsa");
Would this be hard to crack assuming that the one who cracks does not know
the characters that are used to generate the $authstring?
Maybe someone have experience with this? Or just a comment?
Best regards,
Peter Lauri
www.lauri.se <http://www.lauri.se/> - personal web site
www.dwsasia.com <http://www.dwsasia.com/> - company web site
--
PHP General Mailing List (http://www.php.net/)
To unsubscribe, visit: http://www.php.net/unsub.php
