
confidentiality and specifying how to
include features such as digital signatures
and encryption in SOAP documents
(Stanton, 2003). Although dealing with
many important aspects in defining how
security architecture should work, the
specification fails to address other vital
issues with XML Web Services, such as
fail-over, backup, redundancy, etc.
However, this standard is the basis for a
security infrastructure that is being
developed by the same organisation. This
security infrastructure contains other
developing standards, such as: WS-Policy,
WS-Trust, WS-Privacy,
WS-SecureConversation, WS-Federation,
WS-Authorisation. These standards are now
in a drafting stage and are not yet being used
in security implementations.
• Shortcomings of current standards. With
all the standards being proposed, drafted and
approved, there are many people that feel
that developers will soon face another
hurdle: making all of them work together
smoothly. And still, there are unanswered
questions regarding many issues with
security. How will and outside, potentially
weak, system be integrated within an
already secured environment? How do you
perform a security audit, to test that a XML
Web Services environment is secure, when
it is spread over different networks and
administrative domains? (Yang, 2002) What
happens to its clients when a Web Service is
offline, either for maintenance or due to a
malicious action? How will the applications
using that particular service work around the
problem? How does someone prevent
malicious clients from initiating attacks on a
XML Web Service? There are fears that
XML Web Services, in particular
high-profile ones, will become favourite
targets for attackers. There is no way at the
moment to discover a XML Web Service
dynamically and there is also no way of
identifying web services that perform
identical tasks. It becomes obvious that an
attack on a XML Web Service, which might
very well serve thousands or millions of
clients, would become much more damaging
that present-day attacks on web application
servers, which only have the potential to
take down applications running on that
particular server. To address these particular
issues, related specifically to XML Web
Services, a specific kind of security
architecture would have to be designed.
3 ATTACKS ON WEB SERVICES
The main benefit of the XML Web Services
technology is that it exposes functionality,
sometimes critical functionality through standard
HTTP protocols. XML Web Services traffic usually
flows through port 80, such as regular HTTP traffic.
However, as opposed to HTML and regular web
sites, the functionality exposed by a XML Web
Service is much more sensitive. Firewalls can be
configured to intercept SOAP traffic, but most of
them can only block it. If you allow SOAP traffic
through the firewall, there is no way to interpret the
contents of the SOAP messages outside of the XML
Web Service itself, thus a malicious request can
easily go undetected.
Also, and even more dangerous, XML Web
Services are self-descriptive. The contract (WSDL)
file of the service contains detailed information
about the behaviour of the service, from which an
attacker can know exactly what the vulnerabilities of
the service are. This makes attacks much more likely
to succeed than on regular web applications.
Unencrypted or poorly encrypted, SOAP
messages can easily be intercepted and tampered
with. For these reasons, until now, there are no
big-scale implementations using XML Web
Services. The following types of attacks are most
likely to be performed on XML Web Services:
• Denial of service. This is a common attack
on a regular web server that will take the
server down immediately. An attack on a
XML Web Service will take down the Web
Service and the applications that use it. Such
an attack on a XML Web Service is
obviously much more damaging than on a
web server. The problem is even bigger than
it seems (Yang, 2002), because in the case
of Web Services a firewall cannot always
handle the problem. Due to the fact that a
XML Web Service might work a lot to
satisfy a single request, the service can be
easily overwhelmed by legitimate and not
fake requests. This will still take down the
service and the firewall cannot prevent it
without denying legitimate users from
accessing the service.
• Replay attack. A replay attack consists in
recording legitimate requests to the XML
Web Service and then transmitting them
over and over again to the service. However,
this kind of attack can be detected easier by
analyzing and denying identical requests to
the XML Web Service inside a safety time
interval. Still, this attack, combined with a
ICEIS 2004 - INFORMATION SYSTEMS ANALYSIS AND SPECIFICATION
62