Transcript Document
A Practical Approach to Advanced
Threat Detection and Prevention
Title
Agenda
The Palo Alto Networks approach to threat prevention
Zero-day exploit detection with WildFire and PAN-OS 6.0
The rise of mobile malware and attacks on virtualized infrastructure
WildFire Appliance (WF-500) sizing and deployment
3rd party integration with WildFire
Passive DNS and DNS sinkholing
Advanced threat requires a solution, not point products
Protections
1
•
Reduce the
attack surface
Whitelist applications or block
high-risk apps
•
•
Block known viruses, exploits
•
Block commonly exploited file
types
Known viruses
and exploits
EXE, Java,
.LNK, DLL
High-risk
applications
Failed attempts
Detect the
unknown
2
3
Create
protections
Analysis of all application
traffic
Detection and blocking of C&C via:
•
Bad domains in DNS traffic
•
SSL decryption
•
URLs (PAN-DB)
•
WildFire sandboxing of
exploitive files
•
C&C signatures (anti-spyware)
Client Exploit
Command/Control
HTTP
SSL
Successful spearphishing email
DNS
URL /
C&C
Post-compromise
activity
Using application control against
advanced threats
Example 1: Self-updating malware
Repeated pattern of DNS, HTTP, and unknown traffic
The unknown proved to be the most important traffic
A closer look at the unknown session…
Unknown traffic is
frequently caused by
malware using custom
encryption, proprietary
protocols or file transfers
over raw sockets
Example 2: Data exfiltration over DNS
Unknown traffic traversing the DNS port
HTTP using registered/ephemeral ports
Well, Wireshark thinks it’s DNS, so…
It is essential to
control by
application, rather
than by port
Other examples of DNS tunneling
tcp-over-dns
dns2tcp
Iodine
Heyoka
OzymanDNS
NSTX
Takes advantage of recursive
queries to pass encapsulated
TCP messages to/from a
remote DNS server
What’s new in WildFire™
What’s new in WildFire
Support for additional file types and zero-day exploit detection
0-day Windows malware
0-day exploits
Support for multi-OS analysis
Reporting improvements
•
PAN-OS embedded reports
•
Report incorrect verdict
•
Manual malware submission (WF-500)
•
Static analysis, mutexes, services, register key values, etc.
0-day Android malware
WildFire Subscription in PAN-OS 6.0
WildFire
WildFire analysis of PE analysis
Daily signature feed (TP subscription required)
WildFire logs integrated within PAN-OS
WildFire analysis of all other file types (PDF,
Office, APK*, Java)
30-min signature feed
WildFire API* key
Use of WF-500
*APK analysis and WildFire API not yet available on WF-500
WildFire
Subscription
Malware discovered by WildFire per week
PDF/Office/Java are lower in numbers compared to EXE, but when
they hit, it is bad news!
EXE extremely high in count due to lower barrier to entry and ease of use
of packers
PDF/Office commonly used in targeted spear-phishing emails
Java commonly used in drive-by download exploits
File type
Malware/wk
EXE/DLL
221,000
APK
300
Office
110
Java
50
PDF
50
The emerging mobile malware
landscape
The mobile malware problem
Soft target
Many vulnerabilities on older versions of Android (“Beware of
employees’ cheap Android phones”, NW 2/21/14)
“Users are 3 times more likely to succumb to phishing attacks on
their phones than desktop computes” (Aberdeen Group), and “90%
of respondents would not open a suspicious file on a PC, whereas
only 60% of tablet and 56% of smartphone users would exercise
the same caution” (Symantec study)
Powerful platform
Data on handset at risk, but so is the
rest of the corporate network
Mobile devices are PCs on the
network – any attack launched from a
compromised PC can theoretically be
launched from an Android
Mobile malware in use by APT
First known use of APK attachments in APT spear-phishing emails
from Chinese actor groups
Email sent March 24th 2013 to Uyghur activists
Click the app and…
This is what you see…
While this is stolen…
Contacts (stored both on the
phone and the SIM card)
Call logs
SMS messages
Geo-location
Phone data (phone number,
OS version, phone model,
SDK version)
Attacker’s C2 server
Web-based C2 Control Panel
Remote Desktop
Why focus on APK?
Nearly 100% of all new mobile
malware targets Android
Contributing factors:
Large global market share
Slow rate of OS updates on existing
platforms
Very easy to run arbitrary software
on Android (no jailbreak required)
Many Android app stores with little-
to-no quality control
Source: forbes.com (3/24/2014)
Current popular mobile malware techniques
Coaxing the download
Mobile malware attached to spear-phishing emails to lure an installation
Masquerading as popular apps (sometimes as “free” versions of non-free software)
Abusing user ignorance
Mobile malware asks for many permissions, knowing
user will quickly click-through (similar to SSL clickthrough problem)
Mobile malware asks for the ability to install additional
applications, which is equivalent to giving near-total
permission to the malware
Causing mayhem
Data theft (contacts, email, data)
Espionage (audio/video recording, location)
Financial fraud (banking credential theft, SMS scams)
Detect mobile malware on the network and the endpoint
Palo Alto Networks solution offers three opportunities to detect mobile
malware
Antivirus APK signatures detects the download of known Android
malware over the network
WildFire detects the download of unknown Android malware over the
network
GlobalProtect MSM detects presence of known malware already on the
device
Content
WildFire
TM
GlobalProtect MSM
Unknown APK
upload to WildFire
Detect presence of
known malware on
endpoint
Detect download of
known malware
GlobalProtect Gateway
Detect download of
unknown malware
WildFire Appliance (WF-500)
Enables a private cloud deployment of WildFire
Preferred choice for sensitive networks where files cannot leave the
local network for dynamic analysis
Architecturally equivalent to public cloud deployment
APT Add-on Approach
WildFire Approach
Web Sandbox
WildFire
Manual analysis
Central manager
Email Sandbox
TM
File share Sandbox
WildFire cloud or
appliance
WF-500 Sizing
WildFire Appliance (WF-500) is sized
to meet analysis demands of large
networks
Firewalls analyze millions of sessions
Ingress traffic
Millions
WF-500 statically prescreens most
files
Remainder of files are dynamically
analyzed
Tip for accurate sizing prediction – use
the file blocking profile
All executables, Java, and APK files
are sandboxed
PDF and Office documents are “prescreened” using static analysis
About 10-20% make it to dynamic
analysis
All sessions carrying
file transfers
Known malware
blocked
Unknown files sent
to WildFire
Hundreds
Requires dynamic
analysis
Threats facing virtualized
environments
New Passive DNS Monitoring
Passive DNS sensors collect non-recursive DNS queries performed
by local DNS
Anonymous (no client IPs)
Low data rate (usually up to 1 MB per minute at most)
Builds large database of domain resolution history, including all
resource record types (A, AAAA, MX, NS, TXT, etc)
Malicious domains can be “predicted” based on variety of signals:
NX A or A NX
Shared known bad IP
Shared known bad NS
Name heuristics such as character randomness, domain within a domain,
etc.
Malicious domains added daily to DNS signature set in Anti-spyware
profile
Configuring Passive DNS
Passive DNS is enabled via the anti-spyware profile:
New local DNS sinkholing
Discover and confirm compromised hosts via DNS
Trace back to the actual machine without client DNS visibility
Safely block malicious DNS queries and redirect to sinkhole for intel
collection
Where is
badguy.com?
Compromised host
Malicious
DNS / C2
Local DNS
badguy.com =
10.0.1.201
Command-and-control traffic
Sinkhole
10.0.1.20
1
Integrating network and host indicators
How it works
Clients running
agents
WildFire
Samples
TM
WildFire forensics
(via WildFire API)
4
1
WildFire logs
2
WildFire logs
(via device mgmt API)
3
Bit9 Central Manager
5
• Interrogations using host-based
indicators of compromise
• Whitelist/blacklisting by file hash
Splunk App for Palo Alto Networks
Integrating network and host indicators