(The author is NOT a lawyer!) Chain of Custody refers to the physical, demonstrable, chronological documentation (paper) history, or trail, of the capture or seizure, custody, control points and methods, transfer, storage, check in/out, analysis, and eventual disposition of a piece of “evidence”, whether it is digital or physical. CoC is most often accomplished (in the author’s experience) by noting the time of evidence acquisition, the details of that acquisition, and the actual piece of evidence itself on a standard form used throughout the organization. CoC is maintained by the paper trail, log, form updates, and storage in at least a tamper evident “locker”, which is then behind a locked door. A primary issue in CoC is that if the evidence can be “changed”. The opposing side will be able to challenge the validity of the evidence item and the process used to acquire and store the evidence, thus it is likely to be inadmissible..
Suggestions for Evidence Data
Be self-documenting! Develop a “case” directory and data structure, follow it, and make the naming convention intelligent enough to be useful. For example, name directories YYYYMMDD_CASETYPE_SUBJECTNAME. The case types would be for your organization. For example, “AV” for antivirus, “HR” for a Human Resources case, “ABUSE” from the abuse@ email handle, “EXTATTACK” for external cyber-attack, “AUP” for Acceptable Use Policy Issues (a subset of HR cases where Security identified the issue first), etc.
As you capture data in your case directory, organize it in “Box##” folders. Box folder names can be data sources, user names and then data sources, and other organizational support structures. What is important is that your case notes describe what is in a “Box”, you keep “Boxes” clean, and you avoid mixing data types in “Boxes”. using a self-documenting standard. For example YYYYMMDD_HHMM_SOURCE_TYPE_USER. The SOURCE can be a server name, an application, a workstation name – basically the proper name of the data source. The TYPE is used to explain the type of data captured. For example: 20140202_1244_WEBSENSE_ BLOCKLOG_JSMITH.csv, would be Websense block activity for a particular user called J. Smith collected on 2/2/14 at 12:44 PM.
Name data collection files using a self-documenting standard. For example YYYYMMDD_HHMM_SOURCE_TYPE_USER. The SOURCE can be a server name, an application, a workstation name – basically the proper name of the data source. The TYPE is used to explain the type of data captured. For example: 20140202_1244_WEBSENSE_ BLOCKLOG_JSMITH.csv, would be Websense block activity for a particular user called J. Smith collected on 2/2/14 at 12:44 PM.
Incident responders should read the Federal Rules of Evidence, particularly the article posted below. It is much better to be informed ahead of time.
IR : Identification : Capture and Preserve case information
When working an incident involving a database, the IR team should be sure to understand several key data points about the database itself.
1. What role does the RDBMS provide to the organization, the data it contains, data flow to/from the RDBMS (like an extract), and the sensitivity of that data?
2. Is the data in the RDBMS encrypted, and how secure or tamper evident is the keystore?
3. For authentication: Does the RDBMS utilize localized accounts, centralized accounts, or some mix of authentication models? Is login/logout actually logged (user access)?
4. What is the exposure of the RDBMS and the server(s) it resides on – open services, shares, TCP/UDP ports, trusted authentication?
Microsoft SQL Server Specific Points
1. Presence of database tools on DMZ assets and most, not all servers/systems can be suspicious. For example sqlping found on a DMZ server is of concern, as sqlping is a SQL server scan tool.
2. Look for “output” or “extract” files found on SQL servers. It is likely normal for some output/extracts, but files like “myfile1.txt” or “tableout.csv”. Files that have unexplainable names can be suspicious.
3. By default, members of the “Administrators” group have elevated access to the RDBMS; this isn’t necessary, and should be avoided. Attackers can dump the SAM database using tools like pwdump7, and then work on cracking the hashes.
4. Authentication model and login auditing is configured on the Server Properties page (use the Enterprise Manager utility).
5. Incident response scripts can be created with “sqlcmd.exe”. You can script up select statements, and then package them in WFT!
Filesystem and Registry Notes
The version of SQL server affects, or defines, the default options, logging, and encryption level. The log file can contain login auditing, the method for user authentication (Windows/Mixed), startup information, version information, and other fact data about the instances. A new log is created each time SQL server is started. Up to 6 prior logs are stored in the \LOG\ directory.
Error Log: By default, the error log is located at Program Files\Microsoft SQL Server\MSSQL.n\MSSQL\LOG\ERRORLOG and ERRORLOG.n files.
Version: HKLM\Software\Microsoft\MSSQLServer\ MSSQLServer\CurrentVersion
Instances: HKEY_LOCAL_MACHINE\SOFTWARE\Microsoft\Microsoft SQL Server\Inst2\MSSQLServer\CurrentVersion <<< Inst2 is the instance identifier; there may be multiple instances on the server.
Table 51 File Extension Types
MDF Primary DB; User and objects
NDF Secondary DB; stores data so the database can be spread across several volumes
LDF Transaction log; will store transient data like insert/update/delete; supports rollback for recovery and commit operations once a transaction completes. Tran log entries are registered with a Server Process ID (SPID), which tracks a given session.
TRC Trace file; will contain DDL commands like create, alter, truncate, and delete.
CSV Commonly used for comma delimited exports.
SQL Commonly used for Structured Query Language (SQL) commands.
Copyright © Blue Team Handbook. All rights reserved.
These headers are in the book all together, turned on their side for more readability. They are the same as posted on the NMAP web site, here.