| Relying solely on
reputation analysis is no longer enough to efficiently prevent today’s
sophisticated spam attacks. The industry-leading Predictive Sender
Profiling probes deeper into sent email to identify bad sender behavior
and block identity obfuscation techniques, despite a sender’s lack
of prior spamming history. As the next generation of anti-spam technology,
Predictive Sender Profiling reinforces the superior 95 percent spam
accuracy rate of the Barracuda Spam Firewall.
Reputation vs. Profiling Techniques
Traditionally, the purpose of reputation techniques is to
combat spammers by profiling the sender’s history. Barracuda
Networks utilizes a two-fold approach in determining an email
sender’s reputation: Barracuda Reputation Analysis and Intent
Analysis. Both Reputation and Intent Analysis, like many traditional
reputation techniques, enable the Barracuda Spam Firewall to
block spam efficiently by doing a simple database lookup.
However, as spammers become more organized and more creative
in their tactics, they have resorted to obfuscating their identities
more systematically, rendering reputation data less effective
on its own. Blocking these new forms of spam email requires
the use of techniques that can profile the behavior of the sender
and identify any uncharacteristic activity. Profiling techniques
such as Barracuda Networks Predictive Sender Profiling are designed
to look beyond the apparent reputation of the sender and dig
deeper into the campaign itself to identify anomalous activity.
Reputation Alone Falls Short Against Botnets and Zombies
Sender identity obfuscation techniques often involve spammers
taking control of networks of computers infected with malware
(also called “botnets”), and sending email from diverse sources
throughout the Internet. In doing so, the spammer effectively
hides their own identity from traditional reputation checks
that profile sender network addresses.
For example, in illustrations A, B, and C below, the spammer
attempts to hide their identity by sending out virtually the
same message from different addresses around the world. In illustration
A, the message is detected as originating from an IP address
in Germany, One day later, in illustration B the same message
is picked up in as coming from the UK and by the third day of
the campaign, Barracuda Central had identified the message again,
this time coming from Spain. Clearly, in this example, the spammer
had overtaken a series of computers (botnets) and used them
for this particular campaign pushing Viagra and Cialis.
In addition to sending from different IP addresses, these
sample emails all used different embedded URLs in an attempt
to bypass Intent Analysis. In Illustration A, the URL points
to http://joecalvin.info, in Illustration B, the URL points
to http://www.wwuau.info, and in Illustration C, the URL points
to http://shjindaio.info. Just as botnets have enabled spammers
to send from many sender IP addresses, cheap domain registrations
have enabled spammers to create new domain identities quickly
and inexpensively.

Illustration A: IP Address: 84.163.90.168
(Deutsche Telekom, Germany)

Illustration B: IP Address: 84.13.58.219
(Opal Telecom, UK)

Illustration C: 217.125.88.118 (Telefonica-Data-Espana,
Spain)
Despite the inability to utilize traditional reputation techniques
on these emails, the Barracuda Spam Firewall, blocked these
messages by profiling the sender’s behavior and predicting new
instances of this email. In this case, the profiled behavior
was derived from the need to provide domain name services (DNS)
for all of the new domains. By recognizing that the spammer
configured all of the new domains with the similar DNS settings
as their known spam domains, the Barracuda Spam Firewall was
able to block all instances of these emails using its Real-time
Intent Analysis capabilities.
Hiding Behind the “Good Guy”
By registering new domains or by redirecting to spam Web
domains through reputable blogs, free Web site providers, or
URL redirection services, spammers have also learned to hide
their identity from traditional reputation checks that profile
spam Web domains.
Illustrations D and E below show two separate spamming campaigns
that were recently detected by Barracuda Central in which the
spammers attempt to hide their identity by using URLs referencing
reputable Web domains, Geocities and Blogspot. Often these URLs
contain either redirections or simple Web links to known spammer
Web sites.

Illustration D: Geocities redirect to sexdatesearch.com
– known spammer

Illustration E: Blogspot redirect to known
spammer IP (211.93.46.38)
Despite these attempts to hide behind a “good” identity,
the Barracuda Spam Firewall profiled this campaign behavior
of placing redirections or Web links to known spam sites behind
popular Web providers. The Barracuda Spam Firewall was able
to block these messages through Multi-level Intent Analysis
by following the embedded URLs as a Web browser would and inspecting
the resulting contents.
Sample Behaviors and Countermeasures
When spammers obfuscate their identities, the Barracuda Spam
Firewall can use Predictive Sender Profiling the Barracuda Spam
Firewall to identify behaviors of all senders and apply the
applicable Barracuda Spam Firewall defense tactic. Examples
include:
|
Sample Behaviors: |
Countermeasures: |
| Sending
too many emails from a single network address -
Automated spam software can be used to send large amounts
of email from a single email server. |
Rate Control
- To protect the email infrastructure from these
flood-based attacks, the Barracuda Spam Firewall counts
the number of incoming connections from a particular
IP address and throttles the connections once a particular
threshold is exceeded. |
| Attempting
to send to too many invalid recipients - Many spammers
attack email infrastructures by harvesting email addresses. |
Recipient
Verification - The Barracuda Spam Firewall automatically
rejects SMTP connection attempts from email senders
that attempt to send to too many invalid recipients,
a behavior indicative of directory harvest or dictionary
attacks. |
| Registering
new domains for spam campaigns - Because registering
new domain names is fast and inexpensive, many spammers
switch domain names used in a campaign. |
Real-time
Intent Analysis - Used for new domain names that
may come into use, real-time intent analysis involves
performing DNS lookups and comparing DNS configuration
of new domains against the DNS configurations of known
spammer domains. |
| Using free
Internet services to redirect to known spam domains
- Use of free Web sites to redirect to known spammer
Web sites is a growing practice used by spammers to
hide or obfuscate their identity from mail scanning
techniques such as Intent Analysis. |
Multilevel
Intent Analysis - Multilevel intent analysis involves
inspecting the results of Web queries to URLs of well-known
free Web sites for redirections to known spammer sites. |
|