BTA

About BTA

BTA is an open-source Active Directory security audit framework. Its goal is to help auditors harvest the information they need to answer such questions as:

• Who has rights over a given object (computer, user account, etc.) ?
• Who can read a given mailbox ?
• Which are the accounts with domain admin rights ?
• Who has extended rights (userForceChangePassword, SendAs, etc.) ?
• What are the changes done on an AD between two points in time ?

The framework is made of

• an importer able to translate a ntds.dit file, containing all the AD data, into a database
• tools to query the database
  • AD miner framework
  • AD diff utility
  • small utilities (list of databases, etc.)

The comprehensive set of attributes are imported and can be querried including all schema extensions (Exchange, Sharepoint, etc.).

Each question can be crystallized by an AD expert as a miner, so that it can be used during all audits without doing the hard work again.

Installing BTA

Quick install

Dependencies:

• mongodb (apt-get install mongodb-server)
• python headers (apt-get install python-dev)
• libesedb 2012 alpha version [1]
  • https://bitbucket.org/iwseclabs/bta/downloads/libesedb-alpha-20120102.tar.gz SHA-1 2aba34b1c9703a28b8913019044efa3ff53428fb
  • Build it. Install it or have LD_LIBRARY_PATH point to it when using btaimport

Installation:

• pip install bta

From sources

Dependencies:

• mongodb (apt-get install mongodb-server python-pymongo)
• libesedb 2012 alpha version [1]
  • https://bitbucket.org/iwseclabs/bta/downloads/libesedb-alpha-20120102.tar.gz SHA-1 2aba34b1c9703a28b8913019044efa3ff53428fb
  • Build it. Install it or have LD_LIBRARY_PATH point to it when using btaimport
• openpyxl (>= 2.0.2)

Installation:

• python setup.py install

[1]	(1, 2) support for so called long values is a work in progress in the latest versions and is not working well yet. It is working correctly with 2012 alpha version.

Active Directory Security Analysis

Goal:

• Clean an AD or an AD forest, looking for
  • bad practices
  • forgotten entries
  • backdoors
  • recompromissions
• BTA is an operationnal tool, ought to be
  • deterministic, reliable
  • running a well established procedure

Protocol

Audit steps:

1. Extract the ntds.dit file
2. Import the ntds.dit file in a database
3. Look for control points in the database

Extraction

Here is a way to backup NTDS.dit file for a domain controller which is running on Windows 2008. See [2] for more information or for Windows 2003 method.

ntdsutil
activate instance ntds
ifm
create full c:{\bs}NTDS_saved
quit
quit

[2]	https://www.sstic.org/2012/presentation/audit_ace_active_directory/

Importing

• ntds.dit is unusable as-is.
• one ntds.dit is imported into one MongoDB database
• ability to import several ntds.dit in parallel

Examples:

btaimport -C ::mydb /path/to/ntds.dit
btaimport /path/to/*.dit  --multi             \
  --C-from-filename                         \
     "::%s" "basename rmext 'DB' swap plus"

Analysing

• Querying the database
  • analysing control points of a database: btaminer
  • analysing differences between 2 bases: btadiff

Analysing control points

• miners crystallize expertise
  • list of admin accounts
  • list of accounts with extended rights
  • list of accounts with password errors
  • list of various timelines

btaminer -t ReST -C ::AD1 Schema --timelineCS created

Analysis by miner [Schema]
==========================

+---------------+-----------------------+
| Date          | Affected class schema |
+===============+=======================+
| 2009-02-11 18 | 234                   |
| 2011-12-20 00 | 267                   |
| 2011-12-22 14 | 3                     |
| 2011-12-23 18 | 46                    |
+---------------+-----------------------+

Analysing differences

• diff
  • diff (naive for the moment) between 2 imports at different points in time
  • noise filtering

$ btadiff --CA ::ADclean --CB ::ADbackdoor --ignore-defaults
===============
Starting diffing sd_table
---------------
AB,101: [] *sd_refcount['14'=>'15']
AB,108: [] *sd_refcount['39'=>'41']
A ,229: []
A ,372: []
AB,423: [] *sd_refcount['3'=>'2']
 B,424: []
 B,425: []
 B,428: []
---------------
Table [sd_table]: 160 records checked, 2 disappeared, 3 appeared, 3 changed
===============
[...]

===============
Starting diffing datatable
---------------
AB,3586: [DC001] *logonCount['116'=>'117'], *lastLogon['130052518207794051L'=>'130052535716737649L']
AB,3639: [RID Set] *rIDNextRID['1153'=>'1154']
AB,8784: [A:[gc]/B:[gc  DEL:346bf199-8567-4375-ac15-79ec4b42b270]] +isDeleted,
         *name["u'gc'"=>"u'gc\\nDEL:346bf199-8"], *dc["u'gc'"=>"u'gc\\nDEL:346bf199-8"]
AB,8785: [A:[DomainDnsZones]/B:[DomainDnsZones  DEL:58b2962b-708c-4c93-99ff-0b7e163131f9]]
         +isDeleted, *name["u'DomainDnsZones'"=>"u'DomainDnsZones\\nDE"],
         *dc["u'DomainDnsZones'"=>"u'DomainDnsZones\\nDE"]
AB,8786: [A:[ForestDnsZones]/B:[ForestDnsZones  DEL:87f7d8a2-4d05-48d0-8283-9ab084584470]]
         +isDeleted, *name["u'ForestDnsZones'"=>"u'ForestDnsZones\\nDE"],
         *dc["u'ForestDnsZones'"=>"u'ForestDnsZones\\nDE"]
 B,8789: [snorky insomnihack]
 B,8790: [gc]
 B,8791: [DomainDnsZones]
 B,8792: [ForestDnsZones]
---------------
Table [datatable]: 7636 records checked, 0 disappeared, 4 appeared, 5 changed
===============

Other features

• can give reports in different formats:
  • live dump
  • ReST document
  • zipped tree of CSV files
• audit log of writings in a database
• table consistency checks before mining

Authors

• Airbus Group CERT
• Airbus Group Innovations
• Airbus DS CyberSecurity