This provides a quick introduction to the NTFS file system. The terms used are different then with other file systems. For a full overview of the file system, refer to "File System Forensic Analysis".
The "Inside Windows 2000"book by Solomon and Russinovich also provides a high-level overview and details of the file system structures can be found at the Linux NTFS Source Forge project:
Details on how TSK implemented NTFS can be found in the NTFS Implementation Notes
The Master File Table (MFT) contains entries that describe all system files, user files, and directories. The MFT even contains an entry (#0) that describes the MFT itself, which is how we determine its current size. Other system files in the MFT include the Root Directory (#5), the cluster allocation map, Security Descriptors, and the journal.
Each MFT entry is given a number (similar to inode numbers in UNIX). The user files and directories start at MFT #25. The MFT entry contains a list of attributes. Example attributes include "Standard Information" which stores data such as MAC times, "File Name" which stores the file or directories name(s), $DATA which stores the actual file content, or "Index Alloc" and "Index Root" which contain directory contents stored in a B-Tree.
Each type of attribute is given a numerical value and more than one instance of a type can exist for a file. The "id" value for each attribute allows one to specify an instance. A given file can have more than one "$Data" attribute, which is a method that can be used to hide data from an investigator. To get a mapping of attribute type values to name, use the 'fsstat' command. It displays the contents of the $AttrDef system file.
Each attribute has a header and a value and an attribute is either resident or non-resident. A resident attribute has both the header and the content value stored in the MFT entry. This only works for attributes with a small value (the file name for example). For larger attributes, the header is stored in the MFT entry and the content value is stored in Clusters in the data area. A Cluster in NTFS is the same as FAT, it is a consecutive group of sectors. If a file has too many different attributes, an "Attribute List" is used that stores the other attribute headers in additional MFT entries.
Files in NTFS typically have the following attributes:
- $STANDARD_INFORMATION (#16): Contains MAC times, security ID, Owners ID, permissions in DOS format, and quota data.
- $FILE_NAME (#48): Contains the file name in UNICODE, as well as additional MAC times, and the MFT entry of the parent directory.
- $OBJECT_ID (#64): Identifiers regarding the files original Object ID, its birth Volume ID, and Domain ID.
- $DATA (#128): The raw content data of the file.
When a file is deleted, the IN_USE flag is cleared from the MFT entry, but the attribute contents still exist.
Directories in NTFS are indexed to make finding a specific entry in them faster. By default, they are stored in a B-Tree sorted in alphabetical order. There are two attributes that describe the B-Tree contents. Directories in NTFS typically have the following attributes:
* $STANDARD_INFORMATION (#16): See above * $FILE_NAME (#48): See above * $OBJECT_ID (#64): See above * $INDEX_ROOT (#144): The root of the B-Tree. The $INDEX_ROOT value is one more more "Index Entry" structures that each describe a file or directory. The "Index Entry" structure contains a copy of the "$FILE_NAME" attribute for the file or sub-directory. * $INDEX_ALLOCATION (#160): The sub-nodes of the B-Tree. For small directories, this attribute will not exist and all information will be saved in the $INDEX_ROOT structure. The content of this attribute is one or more "Index Buffers". Each "Index Buffer" contains one or more "Index Entry" structures, which are the same ones found in the $INDEX_ROOT. * $BITMAP (#176): This describes which structures in the B-Tree are being used.
When files are deleted from a directory, the tree node is removed and the tree is resorted. Therefore, the "Index Entry" for the deleted file maybe written over when the tree is resorted. This is different than what is usually seen with UNIX and FAT file systems, which always have the original name and structure until a new file is created. Also, when the tree is resorted, a file that is on the bottom of the tree can be moved up and a deleted file name will exist for the original location (even though it was never deleted by a user).