Aloha, Thad.

To clarify this point about SSL and https:

> User B can type in https://siteC to go directly to
> site C using SSL if site C supports SSL and user B
> has personal knowledge of the correct FQDN for the
> SSL server of site C -- in this case the MITM attack
> fails, even though the MITM is still ITM, when the
> MITM is unable to use cryptanalysis to break the SSL
> encryption.

The MITM is still ITM by virtue of the DNS hijacking.
But, the MITM has no choice but to relay SSL packets
unmodified to site C if the MITM wants to avoid being
detected by user B. Only site C has a copy of the
private key that corresponds to site C's encryption
certificate (public key and server FQDN signed by CA)
therefore only site C can satisfy user B's client and
authenticate as site C. The MITM still proxies user
B's access to site C, but without the ability to
eavesdrop or masquerade so there is no damage done.

Aloha & Mahalo,

Jason Coombs
[EMAIL PROTECTED]

-----Original Message-----
From: Jason Coombs [mailto:[EMAIL PROTECTED]]
Sent: Wednesday, April 24, 2002 9:56 AM
To: Thad Horak
Cc: [EMAIL PROTECTED]
Subject: RE: Session Hijacking


Aloha, Thad.

Yes. One technique is for attacker A to hijack DNS of
site C such that the DNS hijacking impacts user B's
name resolution of site C's domain. Attacker A is able
to direct user B to masquerading server D which then
functions as an unauthorized proxy for user B's access
to site C. DNSSEC will make it harder to hijack DNS of
site C when deployed widely (see
http://www.ietf.org/html.charters/dnsext-charter.html)

User B can type in https://siteC to go directly to
site C using SSL if site C supports SSL and user B
has personal knowledge of the correct FQDN for the
SSL server of site C -- in this case the MITM attack
fails, even though the MITM is still ITM, when the
MITM is unable to use cryptanalysis to break the SSL
encryption.

If, however, user B types in http://siteC and has no
personal knowledge of the FQDN of site C's SSL server
then attacker A is able to fool user B into clicking
on a hyperlink to SSL server D, which obviously uses
an FQDN other than site C's but user B sees exactly
what they expect to see in the way of a user interface
and Web application services because server D makes
an SSL connection to the authentic site C and plays
the SSL-secured MITM.

Good luck. We all need it.

Jason Coombs
[EMAIL PROTECTED]


-----Original Message-----
From: Thad Horak [mailto:[EMAIL PROTECTED]]
Sent: Tuesday, April 23, 2002 1:08 PM
To: [EMAIL PROTECTED]
Subject: Session Hijacking


All,
 
 A peer recently told me that the a network topology
consisting of internal servers routing traffic through
a firewall to the internet was a security hole since
the session could either be hijacked or be hacked
using a MITM technique.

Example:

Internal_server --> PIX NAT --> Internet partner
 
 I understand the fundamentals behind hijacking and
MITM attacks, but it would seem to me that the only
way that an attacker could pull of this type of an
attack would be to compromise a host on the same
switch/hub that the firewall is on. Is this this a
correct assumption? Can attacker A in California
hijack User B in Ohio shopping on Site C in Florida
without compromising some key piece of equipment in
between B and C first?
 
 My apologies for the long winded question. Thanks in
advance for your insight.
 
 Thad

__________________________________________________
Do You Yahoo!?
Yahoo! Games - play chess, backgammon, pool and more
http://games.yahoo.com/

Reply via email to