Benchmarking Web Application Scanners for YOUR Organization

Benchmarking Web Application Scanners
for YOUR Organization
Dan Cornell!
CTO, Denim Group!
@danielcornell

© Copyright 2012 Denim Group - All Rights Reserved

My Background
•  Dan Cornell, founder and CTO of
Denim Group

•  Software developer by background
(Java, .NET, etc)

•  OWASP San Antonio, Global
Membership Committee

© Copyright 2012 Denim Group - All Rights Reserved 2

Denim Group Background

•  Secure software services and products company
–  Builds secure software
–  Helps organizations assess and mitigate risk of in-house developed and third party
software
–  Provides classroom training and e-Learning so clients can build software securely
•  Software-centric view of application security
–  Application security experts are practicing developers
–  Development pedigree translates to rapport with development managers
–  Business impact: shorter time-to-fix application vulnerabilities
•  Culture of application security innovation and contribution
–  Develops open source tools to help clients mature their software security programs
•  Remediation Resource Center, ThreadFix
–  OWASP national leaders & regular speakers at RSA, SANS, OWASP, ISSA, CSI
–  World class alliance partners accelerate innovation to solve client problems


What Do You Want From a Scanner?
•  Coverage

•  Low False Positives

•  Low False Negatives


Scanner Coverage
•  You can’t test what you can’t see

•  How effective is the scanner’s crawler?

•  How are URLs mapped to functionality?
–  RESTful
–  Parameters

•  Possible issues:
–  Login routines
–  Multi-step processes
–  Anti-CSRF protection


Are You Getting a Good Scan?
Large financial firm: “Our 500 page website is secure because the
scanner did not find any vulnerabilities!”

Me: “Did you teach the scanner to log in so that it can see more than just
the homepage?”

Large financial firm: “…”


Can Your Scanner Do This?
•  Two-step login procedure:
–  Enter username / password (pretty standard)
–  Enter answer to one of several arbitrary questions

•  Challenge was that the parameter indicating the question was
dynamic
–  Question_1, Question_2, Question_3, and so on
–  Makes standard login recording ineffective


It All Started With A Simple Blog Post…
•  Ran into an application with a complicated login procedure
•  Wrote blog post about the toolchain used to solve the problem
–  http://blog.denimgroup.com/denim_group/2012/04/automated-application-scanning-handling-
complicated-logins-with-appscan-and-burp-suite.html
•  Other scanner teams responded:
–  IBM Rational AppScan
•  http://blog.denimgroup.com/denim_group/2012/04/automated-application-scanning-handling-complicated-
logins-with-appscan-only.html
–  HP WebInspect
•  http://blog.denimgroup.com/denim_group/2012/05/handling-challengeresponse-logins-in-hp-
webinspect.html
–  Mavituna Security Netsparker
•  http://blog.denimgroup.com/denim_group/2012/05/handling-challengeresponse-logins-in-mavituna-
netsparker.html
–  NTObjectives NTOSpider
•  http://blog.denimgroup.com/denim_group/2012/05/handling-challengeresponse-logins-in-ntospider.html


Scanner Authentication Scenario Examples
•  Built as a response to the previously-mentioned blog conversation

•  Example implementations of different login routines
–  How can different scanners be configured to successfully scan?

•  GitHub site:
–  https://github.com/denimgroup/authexamples


Did I Get a Good Scan?
•  Scanner training is really important
–  Read the Larry Suto reports…

•  Must sanity-check the results of your scans

•  What URLs were accessed?
–  If only two URLs were accessed on a 500 page site, you probably have a bad scan
–  If 5000 URLs were accessed on a five page site, you probably have a bad scan

•  What vulnerabilities were found and not found?
–  Scan with no vulnerabilities – probably not a good scan
–  Scan with excessive vulnerabilities – possibly a lot of false positives


Low False Positives
•  Reports of vulnerabilities that do not actually exist

•  How “touchy” is the scanner’s testing engine?

•  Why are they bad?
–  Take time to manually review and filter out
–  Can lead to wasted remediation time


Low False Negatives
•  Scanner failing to report vulnerabilities that do exist

•  How effective is the scanner’s testing engine?

•  Why are they bad?
–  You are exposed to risks you do not know about
–  You expect that the scanner would have found certain classes of vulnerabilities

•  What vulnerability classes do you think scanners will find?


Other Benchmarking Efforts
•  Larry Suto’s 2007 and 2010 reports
–  Analyzing the Accuracy and Time Costs of Web Application Security Standards
–  http://ha.ckers.org/files/Accuracy_and_Time_Costs_of_Web_App_Scanners.pdf
–  Vendor reactions were … varied
–  [Ofer Shezaf attended this talk at AppSecEU 2012 and had some great questions
and comments. See his reactions to the latest Larry Suto scanner report here :
http://www.xiom.com/2010/02/09/wafs-are-not-perfect-any-security-tool-perfect ]

•  Shay Chen’s Blog and Site
–  http://sectooladdict.blogspot.com/
–  http://www.sectoolmarket.com/

•  Web Application Vulnerability Scanner Evaluation Project (wavsep)
–  http://code.google.com/p/wavsep/

So I Should Just Buy the Best Scanner, Right?
•  Or the cheapest?

•  Well…
–  What do you mean by “best”?

•  Follow-on questions
–  How well do the scanners work on your organization’s applications?
–  How many false positives are you willing to deal with?
–  What depth and breadth of coverage do you need?


ThreadFix - Overview
•  ThreadFix is a software vulnerability aggregation and management
system that helps organizations aggregate vulnerability data, generate
virtual patches, and interact with software defect tracking systems.

•  Freely available under the Mozilla Public License (MPL)

•  Hosted at Google Code: http://code.google.com/p/threadfix/


What is a Unique Vulnerability?
•  (CWE, Relative URL)
–  Predictable resource location
–  Directory listing misconfiguration

•  (CWE, Relative URL, Injection Point)
–  SQL injection
–  Cross-site Scripting (XSS)

•  Injection points
–  Parameters – GET/POST
–  Cookies
–  Other headers


What Do The Scanner Results Look Like?
•  Usually XML
–  Skipfish uses JSON and gets packaged as a ZIP

•  Scanners have different concepts of what a “vulnerability” is
–  We normalize to the (CWE, location, [injection point]) noted before

•  Look at some example files

•  Several vendors have been really helpful adding additional data to
their APIs and file formats to accommodate requests


Why Common Weakness Enumeration (CWE)?
•  Every tool has their own “spin” on naming vulnerabilities
•  OWASP Top 10 / WASC 24 are helpful but not comprehensive

•  CWE is exhaustive (though a bit sprawling at times)
•  Reasonably well-adopted standard
•  Many tools have mappings to CWE for their results

•  Main site: http://cwe.mitre.org/


Demo
•  Unpack and install ThreadFix
•  Use ThreadFix to normalize and report on the use of multiple scanning
technologies on a given application
•  Import multiple scans and de-duplicate the results

•  These screenshots are based on UNTUNED scans and are NOT
meant to show a real benchmark of these scanners – only the
process


Unzip the ThreadFix Package (like WebGoat!)


Make threadfix.sh Executable


Run ThreadFix Pre-Configured Tomcat Server


Login to ThreadFix (“user” and “password”)


Upload Various Scan Results Files


This Vulnerability Found By Three Scanners


Mark False Positives
(wavsep Uses “FalsePositives” In the URL…)


Summary Report – Found, Not Found, False
Positives (Again – NOT Based on Tuned Scans)


Report By Vulnerability Type


Detail Report Can Be Used To Error-Check
Merge Process


Current Limitations
•  Vulnerability importers are not currently
formally vendor-supported
–  Though a number have helped us test and
refine them (thanks!)
–  After you get a good scan make sure you also
got a good import

•  Summary report should show data by
severity rating
–  Make it easier to focus on vulnerabilities you
probably care more about
–  But you can look at the data by vulnerability
type


Try This At Home, Kids
•  Pick some applications to test
–  Representative sample for your organization
–  Common languages, frameworks

•  Run scans with the targeted scanning technologies
–  Make sure you get good scans: login, other state-based issues
–  If you train the scans (always a good idea) be consistent

•  Import the scans into ThreadFix
–  Make sure you’re happy with the import

•  Run some reports


You Know What Would Make All This Way Easier?
•  Common data standards for scanning
tools!

•  Current efforts:
–  MITRE Software Assurance Findings
Expression Schema (SAFES)
•  http://www.mitre.org/work/tech_papers/
2012/11_3671/
–  OWASP Data Exchange Format Project
•  https://www.owasp.org/index.php/
OWASP_Data_Exchange_Format_Project


Simple Software Vulnerability Language (SSVL)
•  Common way to represent static and dynamic scanner findings
•  Based on our experience building importers for ThreadFix
–  It “works” for real-world applications because we are essentially using it

•  Love to hear feedback
–  Send me a request and I can share the document for editing/annotation

•  Online:
–  https://docs.google.com/document/d/
1H5hWUdj925TtoZ7ZvnfHdFABe7hBCGuZtLUas29yBGI/edit?pli=1
–  Or http://tinyurl.com/cslqv47


Simple Software Vulnerability Language (SSVL)


Questions

Dan Cornell
dan@denimgroup.com
Twitter: @danielcornell

www.denimgroup.com
www.denimgroup.com/threadfix
code.google.com/p/threadfix
(210) 572-4400


Benchmarking Web Application Scanners for YOUR Organization

More Related Content

What's hot (20)

Viewers also liked (6)

Similar to Benchmarking Web Application Scanners for YOUR Organization (20)

More from Denim Group (20)

Recently uploaded (20)

Benchmarking Web Application Scanners for YOUR Organization