Windows Operating System Internals - by David A. Solomon and Mark E. Russinovich with Andreas PolzeWindows Operating System Internals - by David A. Solomon and Mark E. Russinovich with Andreas Polze

Unit OS8: File SystemUnit OS8: File System8.2.8.2. Windows File SystemsWindows File Systems

2Copyright NoticeCopyright Notice 2000-2005 David A. Solomon and Mark Russinovich 2000-2005 David A. Solomon and Mark Russinovich

These materials are part of the These materials are part of the Windows Operating Windows Operating System Internals Curriculum Development Kit,System Internals Curriculum Development Kit, developed by David A. Solomon and Mark E. developed by David A. Solomon and Mark E. Russinovich with Andreas PolzeRussinovich with Andreas Polze

Microsoft has licensed these materials from David Microsoft has licensed these materials from David Solomon Expert Seminars, Inc. for distribution to Solomon Expert Seminars, Inc. for distribution to academic organizations solely for use in academic academic organizations solely for use in academic environments (and not for commercial use)environments (and not for commercial use)

3Roadmap for Section 8.2Roadmap for Section 8.2

File Systems supported by WindowsFile Systems supported by WindowsNTFS Design GoalsNTFS Design GoalsFile System Driver ArchitectureFile System Driver ArchitectureNTFS OperationNTFS OperationWindows File System On-Disk StructureWindows File System On-Disk StructureNTFS File CompressionNTFS File Compression

4Windows File System - Windows File System - TerminologyTerminology

Sectors: Sectors: hardware-addressable blocks on a storage mediumhardware-addressable blocks on a storage mediumTypical sector size on hard disks for x86-based systems is 512 bytesTypical sector size on hard disks for x86-based systems is 512 bytes

File system formats:File system formats:Define the way data is stored on storage mediaDefine the way data is stored on storage mediaImpact a file system features: permissions & security, limitations on file size, Impact a file system features: permissions & security, limitations on file size, support for small/large files/diskssupport for small/large files/disks

Clusters:Clusters:Addressable blocks that many file system formats useAddressable blocks that many file system formats useCluster size is always a multiple of the sector sizeCluster size is always a multiple of the sector sizeCluster size tradeoff: space efficiency vs. access speedCluster size tradeoff: space efficiency vs. access speed

Metadata:Metadata:Data stored on a volume in support of file system format managementData stored on a volume in support of file system format management

Metadata includes the data that defines the placement of files and Metadata includes the data that defines the placement of files and directories on a volume, for exampledirectories on a volume, for exampleTypically not accessible to applicationsTypically not accessible to applications

5Formats Supported by WindowsFormats Supported by Windows

CD-ROM File System (CDFS)CD-ROM File System (CDFS)

Universal Disk Format (UDF)Universal Disk Format (UDF)

File Allocation Table (FAT12, FAT16, and File Allocation Table (FAT12, FAT16, and FAT32)FAT32)

New Technology File System (NTFS)New Technology File System (NTFS)

6CDFSCDFS

CDFS, or, is a relatively simple format that was defined in 1988 as CDFS, or, is a relatively simple format that was defined in 1988 as the read-only formatting standard for CD-ROM media. the read-only formatting standard for CD-ROM media.

Windows 2000 implements ISO 9660-compliant CDFS in Windows 2000 implements ISO 9660-compliant CDFS in \Winnt\System32\Drivers\Cdfs.sys, with long filename support \Winnt\System32\Drivers\Cdfs.sys, with long filename support defined by Level 2 of the ISO 9660 standarddefined by Level 2 of the ISO 9660 standard

Because of its simplicity, the CDFS format has a number of Because of its simplicity, the CDFS format has a number of restrictionsrestrictions

Directory and file names must be fewer than 32 characters longDirectory and file names must be fewer than 32 characters long

Directory trees can be no more than eight levels deepDirectory trees can be no more than eight levels deep

CDFS is considered a legacy format because the industry has CDFS is considered a legacy format because the industry has adopted the Universal Disk Format (UDF) as the standard for read-adopted the Universal Disk Format (UDF) as the standard for read-only mediaonly media

7UDFUDF

OSTA (Optical Storage Technology Association) defined OSTA (Optical Storage Technology Association) defined UDF in 1995 as a format to replace CDFS for magneto-UDF in 1995 as a format to replace CDFS for magneto-optical storage media, mainly DVD-ROMoptical storage media, mainly DVD-ROM

The Windows 2000 UDF file system implementation is ISO The Windows 2000 UDF file system implementation is ISO 13346-compliant and supports UDF versions 1.02 and 1.513346-compliant and supports UDF versions 1.02 and 1.5

UDF file systems have the following traits:UDF file systems have the following traits:Filenames can be 255 characters longFilenames can be 255 characters long

The maximum path length is 1023 charactersThe maximum path length is 1023 characters

Although the UDF format was designed with rewritable Although the UDF format was designed with rewritable media in mind, the Windows 2000 UDF driver media in mind, the Windows 2000 UDF driver (\Winnt\System32\Drivers\Udfs.sys) provides read-only (\Winnt\System32\Drivers\Udfs.sys) provides read-only supportsupport

8FATFAT

FAT (File Allocation Table) file systems are a legacy format that FAT (File Allocation Table) file systems are a legacy format that originated in DOS and Windows 9xoriginated in DOS and Windows 9x

Reasons why Windows supports FAT file systems:Reasons why Windows supports FAT file systems:to enable upgrades from other versions of Windowsto enable upgrades from other versions of Windows

compatibility with other operating systems in multiboot systemscompatibility with other operating systems in multiboot systems

as a floppy disk formatas a floppy disk format

Windows FAT file system driver is implemented in Windows FAT file system driver is implemented in \Winnt\System32\Drivers\Fastfat.sys\Winnt\System32\Drivers\Fastfat.sys

Each FAT format includes a number that indicates the number of Each FAT format includes a number that indicates the number of bits the format uses to identify clusters on a diskbits the format uses to identify clusters on a disk

Boot sector File allocation table 2 (duplicate)File allocation

table 1 Root directory Other directories and all files

FAT format organizationFAT format organization

9FAT12FAT12

FAT12's 12-bit cluster identifier limits a partition FAT12's 12-bit cluster identifier limits a partition to storing a maximum of 2to storing a maximum of 21212 (4096) clusters (4096) clusters

Windows uses cluster sizes from 512 bytes to 8 KB Windows uses cluster sizes from 512 bytes to 8 KB in size, which limits a FAT12 volume size to 32 MBin size, which limits a FAT12 volume size to 32 MB

Windows uses FAT12 as the format for all 5-inch Windows uses FAT12 as the format for all 5-inch floppy disks and 3.5-inch floppy disks, which store floppy disks and 3.5-inch floppy disks, which store up to 1.44 MB of dataup to 1.44 MB of data

10

FAT16FAT16

FAT16, with a 16-bit cluster identifier, can FAT16, with a 16-bit cluster identifier, can address 2address 21616 (65,536) clusters (65,536) clusters

On Windows, FAT16 cluster sizes range from 512 On Windows, FAT16 cluster sizes range from 512 bytes (the sector size) to 64 KB, which limits FAT16 bytes (the sector size) to 64 KB, which limits FAT16 volume sizes to 4 GBvolume sizes to 4 GB

The cluster size Windows uses depends on the size The cluster size Windows uses depends on the size of a volumeof a volume

11

FAT32FAT32FAT32 is the most recently defined FAT-based file FAT32 is the most recently defined FAT-based file system formatsystem format

it's included with Windows 95 OSR2, Windows 98, and Windows it's included with Windows 95 OSR2, Windows 98, and Windows Millennium EditionMillennium Edition

FAT32 uses 32-bit cluster identifiers but reserves the FAT32 uses 32-bit cluster identifiers but reserves the high 4 bits, so in effect it has 28-bit cluster identifiershigh 4 bits, so in effect it has 28-bit cluster identifiers

Because FAT32 cluster sizes can be as large as 32 KB, FAT32 Because FAT32 cluster sizes can be as large as 32 KB, FAT32 has a theoretical ability to address 8 TB volumeshas a theoretical ability to address 8 TB volumesAlthough Windows works with existing FAT32 volumes of larger Although Windows works with existing FAT32 volumes of larger sizes (created in other operating systems), it limits new FAT32 sizes (created in other operating systems), it limits new FAT32 volumes to a maximum of 32 GBvolumes to a maximum of 32 GBFAT32's higher potential cluster numbers let it more efficiently FAT32's higher potential cluster numbers let it more efficiently manage disks than FAT16; it can handle up to 128-MB volumes manage disks than FAT16; it can handle up to 128-MB volumes with 512-byte clusterswith 512-byte clusters

Unlike FAT12 and FAT16, root directory is not fixed size Unlike FAT12 and FAT16, root directory is not fixed size or locationor location

Largest file size on Windows is 4GB (largest on Win9x is 2G)Largest file size on Windows is 4GB (largest on Win9x is 2G)

12

NTFSNTFS

NTFS is the native file system format of WindowsNTFS is the native file system format of Windows

NTFS uses 64-bit cluster indexesNTFS uses 64-bit cluster indexesTheoretical ability to address volumes of up to 16 exabytes (16 billion Theoretical ability to address volumes of up to 16 exabytes (16 billion GB)GB)

Windows 2000 limits the size of an NTFS volume to that addressable Windows 2000 limits the size of an NTFS volume to that addressable with 32-bit clusters, which is 128 TB (using 64-KB clusters)with 32-bit clusters, which is 128 TB (using 64-KB clusters)

Why use NTFS instead of FAT? FAT is simpler, making it faster for Why use NTFS instead of FAT? FAT is simpler, making it faster for some operations, but NTFS supports:some operations, but NTFS supports:

Larger file sizes and disksLarger file sizes and disks

Better performance on large disks, large directories, and small filesBetter performance on large disks, large directories, and small files

ReliabilityReliability

SecuritySecurity

13

CIFS CIFS the Common Internet File Systemthe Common Internet File System

The standard Windows network file systemThe standard Windows network file systemThe file sharing protocol at the heart of CIFS is an updated version of the The file sharing protocol at the heart of CIFS is an updated version of the Server Message Block (SMB) protocolServer Message Block (SMB) protocol

dates back to the mid-1980s dates back to the mid-1980s in 1996/97, Microsoft submitted draft CIFS specifications to the IETF in 1996/97, Microsoft submitted draft CIFS specifications to the IETF

The SMB protocol was originally developed to run over NetBIOS (Network The SMB protocol was originally developed to run over NetBIOS (Network Basic Input Output System) LANsBasic Input Output System) LANs

Until Windows 2000, NetBIOS support was required for SMB transportUntil Windows 2000, NetBIOS support was required for SMB transportThe machine and service names visible in the Windows Network Neighborhood The machine and service names visible in the Windows Network Neighborhood are, basically, NetBIOS addresses (Windows 2000 and later use DNS names)are, basically, NetBIOS addresses (Windows 2000 and later use DNS names)

Windows 3.11 (WfW) introduced: Windows 3.11 (WfW) introduced: service announcement and location system called Browsing service announcement and location system called Browsing The browser service provides the list of available file and print services The browser service provides the list of available file and print services presented in the Network Neighborhoodpresented in the Network NeighborhoodWorkgroup concept was expanded to create NT DomainsWorkgroup concept was expanded to create NT Domains

14

File System Format CompatibilityFile System Format Compatibility

FAT12/FAT16 supported on all Microsoft OSsFAT12/FAT16 supported on all Microsoft OSs

FAT32:FAT32:Only Windows 2000/XP/2003Only Windows 2000/XP/2003

Winternals FAT32 driver for NT4Winternals FAT32 driver for NT4

NTFS:NTFS:Only Windows NT-based OSsOnly Windows NT-based OSs

Winternals NTFSDOS for DOS accessWinternals NTFSDOS for DOS access

Winternals NTFS for Windows 98 for Win9x/Me Winternals NTFS for Windows 98 for Win9x/Me

15

NTFS Design GoalsNTFS Design Goals

Overcome limitations inherent in FAT / HPFSOvercome limitations inherent in FAT / HPFSFAT (File Allocation Table) does not support large disks very FAT (File Allocation Table) does not support large disks very wellwellFAT16 (MS-DOS file system) supports only up to 2FAT16 (MS-DOS file system) supports only up to 21616 clusters clustersand 2 GB disks (with 64 Kb clusters!!)and 2 GB disks (with 64 Kb clusters!!)FAT / root directory represents single point of failureFAT / root directory represents single point of failureNumber of entries in root directory is limitedNumber of entries in root directory is limitedHPFS removed some of FATs limitations, but still did not HPFS removed some of FATs limitations, but still did not support recoverability, security, data redundancy, and fault-support recoverability, security, data redundancy, and fault-tolerancetolerance(later versions of HPFS support up to 2TeraByte disks)(later versions of HPFS support up to 2TeraByte disks)

16

NTFS RecoverabilityNTFS Recoverability

PC disk I/O in the old days: Speed was most importantPC disk I/O in the old days: Speed was most importantNTFS changes this view Reliability counts most:NTFS changes this view Reliability counts most:

I/O operations that alter NTFS structure are implemented as atomic I/O operations that alter NTFS structure are implemented as atomic transactionstransactions

Change directory structure, Change directory structure, extend files, allocate space for new filesextend files, allocate space for new files

Transactions are either completed or rolled backTransactions are either completed or rolled backNTFS uses redundant storage for vital FS informationNTFS uses redundant storage for vital FS information

Contrasts with FAT / HPFS on-disk structures, which have single Contrasts with FAT / HPFS on-disk structures, which have single sectors containing critical file system datasectors containing critical file system dataRead error in these sectors -> volume lostRead error in these sectors -> volume lost

17

NTFS Security NTFS Security and Recoverabilityand Recoverability

NTFS security is derived from NTFS security is derived from Windows Windows object modelobject modelOpen file is implemented as file object;Open file is implemented as file object;security descriptor is stored on disk as part of the filesecurity descriptor is stored on disk as part of the fileNT security system verifies access rights when a process tries to open a handle to NT security system verifies access rights when a process tries to open a handle to any objectany objectAdministrator or file owner may set permissionsAdministrator or file owner may set permissions

NTFS recoverability ensures integrity of FS structureNTFS recoverability ensures integrity of FS structureNo guarantees for complete recovery of user filesNo guarantees for complete recovery of user filesLayered driver model + FTDISK driverLayered driver model + FTDISK driver

Mirroring of data RAID level 1Mirroring of data RAID level 1Striping of data - RAID level 5 (one disk with parity info)Striping of data - RAID level 5 (one disk with parity info)

18

Large Disks and Large FilesLarge Disks and Large Files

Efficient support for large files and disks in NTFSEfficient support for large files and disks in NTFSFAT16:FAT16:

16-bit wide table stores allocation status of disk 16-bit wide table stores allocation status of disk Up to 65.536 clusters per volume (#files !!); adjustable cluster sizeUp to 65.536 clusters per volume (#files !!); adjustable cluster size

FAT32:FAT32:New in since Windows 2000New in since Windows 20004kb clusters on volumes up to 8 GB4kb clusters on volumes up to 8 GBCan relocate root directory / use backup copy of FATCan relocate root directory / use backup copy of FATRoot directory is ordinary cluster chain no limits on #entriesRoot directory is ordinary cluster chain no limits on #entries

HPFS (support dropped in NT 4.0):HPFS (support dropped in NT 4.0):32 bits to enumerate allocation units; maximum file size: 4GB32 bits to enumerate allocation units; maximum file size: 4GBAllocates disk space in terms of physical sectors of 512 bytes;Allocates disk space in terms of physical sectors of 512 bytes;problem with some disks (1024 bit sectors)problem with some disks (1024 bit sectors)

19

Large Disks and Large Files Large Disks and Large Files (contd.)(contd.)

NTFS enumerates cluster with 64-bit numbersNTFS enumerates cluster with 64-bit numbersUp to 2Up to 26464 clusters of up to 64 Kbytes size clusters of up to 64 Kbytes sizeMaximum file size: 2Maximum file size: 26464 bytes bytesCluster size is adjustableCluster size is adjustable

512 bytes on small disks512 bytes on small disksMaximum of 64Kb on large disksMaximum of 64Kb on large disks

Multiple data streamsMultiple data streamsFile info: name, owner, time stamps, type implemented as attributeFile info: name, owner, time stamps, type implemented as attributeEach attribute consists of a stream sequence of bytesEach attribute consists of a stream sequence of bytesDefault data stream has no nameDefault data stream has no nameNew streams can be added: myfile.dat:stream2New streams can be added: myfile.dat:stream2File operations manipulate all streams simultaneouslyFile operations manipulate all streams simultaneously

Used to implementservices for Macintoshin Windows NT Server

20

Other NTFS FeaturesOther NTFS Features

Multiple data streams Multiple data streams

Unicode-based names Unicode-based names

Hard links Hard links

Junctions Junctions

Compression and sparse files Compression and sparse files

Change logging Change logging

Per-user volume quotas Per-user volume quotas

Link tracking Link tracking

Encryption Encryption

POSIX support POSIX support

DefragmentationDefragmentation

21

Multiple Data StreamsMultiple Data Streams

In NTFS, each unit of information associated with a file, In NTFS, each unit of information associated with a file, including its name, its owner, its time stamps, its including its name, its owner, its time stamps, its contents, and so on, is implemented as a file attribute contents, and so on, is implemented as a file attribute (NTFS object attribute)(NTFS object attribute)

Each attribute consists of a single Each attribute consists of a single stream,stream, that is, a that is, a simple sequence of bytessimple sequence of bytes

This generic implementation makes it easy to add more This generic implementation makes it easy to add more attributes (and therefore more streams) to a fileattributes (and therefore more streams) to a file

Because a file's data is "just another attribute" of the file and Because a file's data is "just another attribute" of the file and because new attributes can be added, NTFS files (and file because new attributes can be added, NTFS files (and file directories) can contain multiple data streamsdirectories) can contain multiple data streams

22

Multiple Data StreamsMultiple Data Streams

An NTFS file has one default data stream, which has no An NTFS file has one default data stream, which has no namename

An application can create additional, named data streams and An application can create additional, named data streams and access them by referring to their names. access them by referring to their names.

To avoid altering the Microsoft Windows I/O APIs, which take To avoid altering the Microsoft Windows I/O APIs, which take a string as a filename argument, the name of the data stream a string as a filename argument, the name of the data stream is specified by appending a colon (:) to the filename e.g. is specified by appending a colon (:) to the filename e.g. myfile:stream2myfile:stream2

23

Unicode NamesUnicode Names

Like Windows as a whole, NTFS is fully Unicode Like Windows as a whole, NTFS is fully Unicode enabled, using Unicode characters to store enabled, using Unicode characters to store names of files, directories, and volumesnames of files, directories, and volumes

24

Hard LinksHard Links

A hard link allows multiple paths to refer to the same file A hard link allows multiple paths to refer to the same file or directoryor directory

If you create a hard link named C:\Users\Documents\Spec.doc If you create a hard link named C:\Users\Documents\Spec.doc that refers to the existing file C:\My Documents\Spec.doc, the that refers to the existing file C:\My Documents\Spec.doc, the two paths link to the same on-disk file and you can make two paths link to the same on-disk file and you can make changes to the file using either pathchanges to the file using either path

can create hard links can create hard links with the Windoqs API with the Windoqs API CreateHardLinkCreateHardLink function functionor the or the lnln POSIX function POSIX function

25

JunctionsJunctions

Junctions, also called symbolic links, allow a Junctions, also called symbolic links, allow a directory to redirect file or directory pathname directory to redirect file or directory pathname translation to an alternate directorytranslation to an alternate directory

If the path C:\Drivers is a junction that redirects to If the path C:\Drivers is a junction that redirects to C:\Winnt\System32\Drivers, an application reading C:\Winnt\System32\Drivers, an application reading C:\Drivers\Ntfs.sys actually reads C:\Drivers\Ntfs.sys actually reads C:\Winnt\System\Drivers\Ntfs.sysC:\Winnt\System\Drivers\Ntfs.sys

Junctions are a useful way to lift directories that are Junctions are a useful way to lift directories that are deep in a directory tree to a more convenient depth deep in a directory tree to a more convenient depth without disturbing the original tree's structure or without disturbing the original tree's structure or contentscontents

26

JunctionsJunctions

You can create junctions with the You can create junctions with the junctionjunction tool tool from Sysinternals or the from Sysinternals or the linkdlinkd tool from the tool from the Resource KitsResource Kits

27

Change LoggingChange Logging

Many types of applications, such as incremental Many types of applications, such as incremental backup utilities, need to monitor a volume for backup utilities, need to monitor a volume for changeschanges

An obvious way to watch for changes is to An obvious way to watch for changes is to perform a full scanperform a full scan

Very performance inefficientVery performance inefficient

There is a way for an application to wait on a There is a way for an application to wait on a directory and be told of notificationsdirectory and be told of notifications

An application can miss changes since it must An application can miss changes since it must specify a buffer to hold themspecify a buffer to hold them

28

Change LoggingChange Logging

With Windows 2000, NTFS introduces With Windows 2000, NTFS introduces the change log, which is a sparse the change log, which is a sparse metadata file that records file system metadata file that records file system events (not enabled by default)events (not enabled by default)

As the file exceeds its maximum on-disk As the file exceeds its maximum on-disk size, NTFS frees the disk space for the size, NTFS frees the disk space for the oldest portions marking them emptyoldest portions marking them empty

An application uses Win32 APIs to read An application uses Win32 APIs to read eventsevents

The log file is shared, and generally large The log file is shared, and generally large enough that an application wont miss enough that an application wont miss changes even during heavy file system changes even during heavy file system activityactivity

29

Per-User Volume QuotasPer-User Volume Quotas

NTFS quota-management support allows for per-user NTFS quota-management support allows for per-user specification of quota enforcementspecification of quota enforcement

Can be configured to log an event indicating the occurrence to Can be configured to log an event indicating the occurrence to the system Event Log if a user surpasses his warning limitthe system Event Log if a user surpasses his warning limit

If a user attempts to use more volume storage then her quota If a user attempts to use more volume storage then her quota limit permits, NTFS can log an event to the system Event Log limit permits, NTFS can log an event to the system Event Log and fail the application file I/O that would have caused the and fail the application file I/O that would have caused the quota violation with a "disk full" error codequota violation with a "disk full" error code

User disk space is tracked on a per-volume basis by User disk space is tracked on a per-volume basis by summing the summing the logicallogical sizes of all the files and directories sizes of all the files and directories that have the user as the owner in their security that have the user as the owner in their security descriptorsdescriptors

30

Link TrackingLink Tracking

Several types of symbolic file links are used by layered applicationsSeveral types of symbolic file links are used by layered applicationsShell shortcuts allow users to place files in their shell namespace (on Shell shortcuts allow users to place files in their shell namespace (on their desktop, for example) that link to files located in the file system their desktop, for example) that link to files located in the file system namespacenamespace

Object linking and embedding (OLE) links allow documents from one Object linking and embedding (OLE) links allow documents from one application to be transparently embedded in the documents of other application to be transparently embedded in the documents of other applicationsapplications

In the past, these links were difficult to manageIn the past, these links were difficult to manageIf someone moved a link source (what a link points to), the link brokeIf someone moved a link source (what a link points to), the link broke

Windows now has a link-tracking service, TrkWks (it runs in Windows now has a link-tracking service, TrkWks (it runs in services.exe), tags link sources with a unique object ID services.exe), tags link sources with a unique object ID

NTFS can return the name of a file given a link, so if the link moves NTFS can return the name of a file given a link, so if the link moves the service can query each of a systems volume for the object IDthe service can query each of a systems volume for the object ID

A distributed link-tracking service, TrkSvr, works to track link source A distributed link-tracking service, TrkSvr, works to track link source movement across systemsmovement across systems

31

EncryptionEncryption

While NTFS implements security for files and directories, the While NTFS implements security for files and directories, the security is ineffective if the physical security of the computer is security is ineffective if the physical security of the computer is compromisedcompromised

Can install a parallel copy of WindowsCan install a parallel copy of Windows

NTFSDOSNTFSDOS

Encrypting File System (EFS)Encrypting File System (EFS)Like compression, its operation is transparentLike compression, its operation is transparent

Also like compression, encryption is a file and directory attributeAlso like compression, encryption is a file and directory attribute

Files that are encrypted can be accessed only by using the private Files that are encrypted can be accessed only by using the private key of an account's EFS private/public key pair, and private keys are key of an account's EFS private/public key pair, and private keys are locked using an account's passwordlocked using an account's password

While you might think that its implemented as a file system filter While you might think that its implemented as a file system filter driver, its a driver thats tightly connected to NTFSdriver, its a driver thats tightly connected to NTFS

32

POSIX SupportPOSIX Support

POSIX support requires two file system features:POSIX support requires two file system features:Primary group in security descriptorPrimary group in security descriptor

Case-sensitive namesCase-sensitive names

33

DefragmentationDefragmentation

Fragmentation: A file is fragmented if its data Fragmentation: A file is fragmented if its data occupies discontiguous clustersoccupies discontiguous clusters

34

DefragmentationDefragmentationA common myth is that NTFS doesnt fragment, but it A common myth is that NTFS doesnt fragment, but it doesdoes

Defragmentation APIs have been present since NT 4Defragmentation APIs have been present since NT 4

Windows 2000 introduced a non-schedulable graphical Windows 2000 introduced a non-schedulable graphical defragmenterdefragmenter

A command line interface was added in Windows XPA command line interface was added in Windows XP

35

Compression and Sparse FilesCompression and Sparse Files

NTFS supports transparent compression of filesNTFS supports transparent compression of filesWhen a directories is marked compressed it means any files When a directories is marked compressed it means any files or subdirectories are marked compressedor subdirectories are marked compressed

Compression is performed on 16-cluster blocks of a fileCompression is performed on 16-cluster blocks of a file

Use Explorer or the compact tool to compress files (compact Use Explorer or the compact tool to compress files (compact shows compression rations for compressed files)shows compression rations for compressed files)

Sparse files are an application-controlled form of Sparse files are an application-controlled form of compression that define parts of a file as empty those compression that define parts of a file as empty those areas dont occupy any disk spaceareas dont occupy any disk space

Applications use Windows APIs to define empty areasApplications use Windows APIs to define empty areas

36

NTFS File System DriverNTFS File System Driver

Cache manager

Virtual memorymanager

I/O manager

NTFS driver

Fault tolerant driver

Disk driver

Access the mappedfile or flush the cache

Flush thelog file

Write thecache

Log fileservice

Log the transactionRead/write the file

Load datafrom disk

intomemory

Read/writea mirroredor stripedvolume

Read/writethe disk

37

Components related to NTFSComponents related to NTFSCache ManagerCache Manager

System wide caching System wide caching for NTFS and other file systems driversfor NTFS and other file systems driversIncluding network file system drivers (server and redirectors)Including network file system drivers (server and redirectors)

Cached files are mapped into virtual memoryCached files are mapped into virtual memorySpecialized interface from Cache Manager to NT virtual memory Specialized interface from Cache Manager to NT virtual memory managermanagerMemory manager calls NTFS to access disk driver and obtain fileMemory manager calls NTFS to access disk driver and obtain file

Log File ServiceLog File Service2 copies of transaction logs2 copies of transaction logsTransaction log is flushed to disk before write-data is sent to diskTransaction log is flushed to disk before write-data is sent to diskCache manager performs actual flush operationCache manager performs actual flush operation

38

NTFS & File ObjectsNTFS & File Objects

File object

Handle tableProcess

File objectData

attribute

User-definedattribute

Streamcontrolblocks

Filecontrolblock

Master filetable

NTFS datastructures

(used to managethe on-diskstructure) NTFS

database(on disk)

Objectmanager

datastructures

App accesses files asNT objects by handles.Object Manager and securitysubsystem verify access rights

39

NTFS On-Disk StructureNTFS On-Disk Structure

Volumes correspond to logical partitions on diskVolumes correspond to logical partitions on diskFault tolerant volumes may span multiple disksFault tolerant volumes may span multiple disks

Windows Windows 2000 2000 Disk Administrator utilityDisk Administrator utility

Volume consists of series of files + unallocated spaceVolume consists of series of files + unallocated spaceFAT volume: some areas specially formatted for file systemFAT volume: some areas specially formatted for file systemNTFS volume: all data are stored as ordinary filesNTFS volume: all data are stored as ordinary files

NTFS refers internally to clustersNTFS refers internally to clustersCluster factor: #sectors/cluster; varies with volume size; Cluster factor: #sectors/cluster; varies with volume size; (integral number of physical sectors; always a power of 2)(integral number of physical sectors; always a power of 2)

Logical Cluster Numbers (LCNs): Logical Cluster Numbers (LCNs): refer to physical locationrefer to physical locationLCNs are contiguous enumeration of all clusters on a volumeLCNs are contiguous enumeration of all clusters on a volume

40

NTFS Cluster SizeNTFS Cluster Size

Default cluster size is disk-size dependentDefault cluster size is disk-size dependent512 bytes for small disks (up to 512 MB)512 bytes for small disks (up to 512 MB)1 KB for disks up to 1 GB1 KB for disks up to 1 GB2 KB for disks between 1 and 2 GB2 KB for disks between 1 and 2 GB4 KB for disks larger than 2 GB4 KB for disks larger than 2 GB

Tradeoff: disk fragmentation versus wasted spaceTradeoff: disk fragmentation versus wasted spaceNTFS refers to physical locations via LCNsNTFS refers to physical locations via LCNs

Physical cluster = LCN * cluster-factorPhysical cluster = LCN * cluster-factor

Virtual Cluster Numbers (VCNs): Virtual Cluster Numbers (VCNs): Enumerates clusters belonging to a file; mapped to LCNsEnumerates clusters belonging to a file; mapped to LCNsLCNs are not necessarily physically contiguousLCNs are not necessarily physically contiguous

41

Master File TableMaster File TableAll data stored on a volume is contained All data stored on a volume is contained in a file in a file MFT: Heart of NTFS volume structureMFT: Heart of NTFS volume structure

Implemented as array of file recordsImplemented as array of file recordsOne row for each file on the volumeOne row for each file on the volume(including one row for MFT itself)(including one row for MFT itself)Metadata files store file system structure Metadata files store file system structure informationinformation(hidden files; $MFT; $Volume...)(hidden files; $MFT; $Volume...)More than one MFT record for highly More than one MFT record for highly fragmented filesfragmented filesNfi.exe Utility from OEM Support Tools Nfi.exe Utility from OEM Support Tools allows to dump MFT content allows to dump MFT content (see (see support.microsoft.com/support/support.microsoft.com/support/kb/articles/Q253/0/66.aspkb/articles/Q253/0/66.asp))

MFT

MFT copy (partial)

Log file

Volume file

Attribute def. table

Root directory

Bitmap file

Boot file

Bad cluster file

User files and dirs.

...

NTFSmetadata

file

42

NTFS operationNTFS operation

Mounting a volumeMounting a volume1.1. NTFS looks in boot file for physical address of MFT ($MFT)NTFS looks in boot file for physical address of MFT ($MFT)2.2. 2nd entry in MFT points to copy of MFT ($MFTMirr)2nd entry in MFT points to copy of MFT ($MFTMirr)

used to locate metadata files if MFT is corruptedused to locate metadata files if MFT is corrupted

1.1. MFT entry in MFT contains VCN-to-LCN mapping infoMFT entry in MFT contains VCN-to-LCN mapping info2.2. NTFS obtains from MFT addresses of metadata filesNTFS obtains from MFT addresses of metadata files

NTFS opens these filesNTFS opens these files

1.1. NTFS performs recovery operationsNTFS performs recovery operations2.2. File system is now ready for user accessFile system is now ready for user access

43

NTFS metadataNTFS metadata

NTFS writes to log file ($LogFile)NTFS writes to log file ($LogFile)Record all commands that change volume structureRecord all commands that change volume structure

Root directory:Root directory:When NTFS tries to open a file, it starts search in the root directoryWhen NTFS tries to open a file, it starts search in the root directoryOnce the file is found, NTFS stores the files MFT file referenceOnce the file is found, NTFS stores the files MFT file referenceSubsequent read/write ops. may access files MFT record directlySubsequent read/write ops. may access files MFT record directly

Bitmap file ($Bitmap):Bitmap file ($Bitmap): stores allocation state volume; each bit represents one clusterstores allocation state volume; each bit represents one cluster

Boot file ($Boot):Boot file ($Boot):Stores bootstrap codeStores bootstrap codeHas to be located at special disk addressHas to be located at special disk addressRepresented as file by NTFS -> file ops. possible (!) (no editing)Represented as file by NTFS -> file ops. possible (!) (no editing)

44

NTFS metadata (contd.)NTFS metadata (contd.)

Bad-cluster file ($BadClus)Bad-cluster file ($BadClus)Records bad spots on the diskRecords bad spots on the disk

Volume file ($Volume)Volume file ($Volume)Contains: volume name, NTFS versionContains: volume name, NTFS versionBit, which indicates whether volume is corruptedBit, which indicates whether volume is corrupted

Attribute Definition Table ($AttrDef)Attribute Definition Table ($AttrDef)Defines attribute types supported on the volumeDefines attribute types supported on the volumeIndicates whether they can be indexed, recovered, Indicates whether they can be indexed, recovered, etc.etc.

45

File Records & File Records & File Reference NumbersFile Reference Numbers

File on NTFS volume is identified by File on NTFS volume is identified by file referencefile referenceFile number == index in MFTFile number == index in MFTSequence number used by NTFS for consistency checking;Sequence number used by NTFS for consistency checking;incremented each time a reference is re-usedincremented each time a reference is re-used

File Records:File Records:File is collection of attribute/value pairs (one of which is data)File is collection of attribute/value pairs (one of which is data)Unnamed data attributeUnnamed data attributeOther attributes: filename, time stamp, security descriptor,...Other attributes: filename, time stamp, security descriptor,...Each file attribute is stored as separate stream of bytes within a fileEach file attribute is stored as separate stream of bytes within a file

Sequence number File number

063 47

46

File Records (contd.)File Records (contd.)

NTFS doesnt read/write files:NTFS doesnt read/write files:It reads/writes attribute streamsIt reads/writes attribute streamsOperations: create, delete, read (byte range), write (byte range)Operations: create, delete, read (byte range), write (byte range)Read/write normally operate on unnamed data attributeRead/write normally operate on unnamed data attribute

FilenameStandard

informationSecurity

descriptor Data

Master File Table

MFT record for a small file

Windows optimization: Security descriptorsare stored in a central file and referenced by each file record (saves disk space)

47

Standard Attributes for NTFS FilesStandard Attributes for NTFS Files

AttributeAttribute DescriptionDescription

Standard Standard informationinformation

File attributes: read-only, archive, etc; time stamps; File attributes: read-only, archive, etc; time stamps; creation/modification time; hard link countcreation/modification time; hard link count

FilenameFilename Name in Unicode characters; multiple filename Name in Unicode characters; multiple filename attributes possible (POSIX links!!); short names for attributes possible (POSIX links!!); short names for access by MS-DOS and 16-bin Win applicationsaccess by MS-DOS and 16-bin Win applications

Security descriptorSecurity descriptor Specifies who owns the file and who can access itSpecifies who owns the file and who can access it

datadata Contents of the file; a file has one default unnamed data Contents of the file; a file has one default unnamed data attribute; directory has no default data attrib.attribute; directory has no default data attrib.

Index root, indexIndex root, index Three attributes used to implement filename allocation, Three attributes used to implement filename allocation, bitmap index for large directories (dirs. only)bitmap index for large directories (dirs. only)

Attribute listAttribute list List of attributes that make up the file and first reference List of attributes that make up the file and first reference of the MFT record in which the attribute is located (for of the MFT record in which the attribute is located (for files which require multiple MFT file records)files which require multiple MFT file records)

48

Attributes (contd.)Attributes (contd.)

Each attribute in a file record has a name and a valueEach attribute in a file record has a name and a valueNTFS identifies attributes:NTFS identifies attributes:

Uppercase name starting with $: $FILENAME, $DATAUppercase name starting with $: $FILENAME, $DATA

Attributes value: Byte streamAttributes value: Byte streamThe filename for $FILENAMEThe filename for $FILENAMEThe data bytes for $DATAThe data bytes for $DATA

Attribute names correspond to numeric typecodesAttribute names correspond to numeric typecodesFile attributes in an MFT record are ordered by typecodesFile attributes in an MFT record are ordered by typecodes

Some attribute types may appear more than once (e.g. Filename)Some attribute types may appear more than once (e.g. Filename)

49

FilenamesFilenamesPOSIX:POSIX:

Case-sensitive, trailing periods & spacesCase-sensitive, trailing periods & spacesNTFS namespace equiv. to POSIX spaceNTFS namespace equiv. to POSIX space

Win32:Win32:Long filenames, unicode namesLong filenames, unicode namesMultiple dots, embedded spaces, beginning Multiple dots, embedded spaces, beginning dotsdots

MS-DOS:MS-DOS:8.3 names, case does not matter8.3 names, case does not matter

NTFS generates MS-DOS names for Win32 NTFS generates MS-DOS names for Win32 files automaticallyfiles automatically

Fully functional aliases for NTFS namesFully functional aliases for NTFS namesStored in same directory as long names; dir /xStored in same directory as long names; dir /x

POSIXsubsystem

Win32subsystem

MS-DOS Win16 clients

Namespaces

50

MS-DOS filenames in NTFSMS-DOS filenames in NTFS

NTFS name and MS-DOS name are stored in same file record and refer to same NTFS name and MS-DOS name are stored in same file record and refer to same filefile

Renaming changes both filenamesRenaming changes both filenamesOpen, read, write, delete work with both names equallyOpen, read, write, delete work with both names equally

POSIX hardlinks are implemented in similar wayPOSIX hardlinks are implemented in similar wayDeleting a file with multiple names only decreases link countDeleting a file with multiple names only decreases link count

Generation of MS-DOS names:Generation of MS-DOS names:1.1. Remove all illegal chars; remove all but one period; truncate to 6 charsRemove all illegal chars; remove all but one period; truncate to 6 chars2.2. Append ~1 to name; truncate extension to 3 chars; all uppercaseAppend ~1 to name; truncate extension to 3 chars; all uppercase3.3. Increment ~1 if filename duplicates an existing name in directoryIncrement ~1 if filename duplicates an existing name in directory

NTFS filenameStandard info MS-DOS filename Security desc. Data

MFT file record with MS-DOS filename attribute

51

Resident & Nonresident AttributesResident & Nonresident Attributes

Small files:Small files:All attributes and values fit into MFTAll attributes and values fit into MFT

Attribute with value in MFT is called residentAttribute with value in MFT is called resident

All attributes start with header (always resident)All attributes start with header (always resident)

Header contains offset to attr. value and length of valueHeader contains offset to attr. value and length of value

NTFS filenameStandard info Security desc. Data

MYFILE.DATRESIDENT

Offset: 8hLength: 14h

header

value

52

Attributes (contd.)Attributes (contd.)

Small directory:Small directory:index root attribute contains index of file references index root attribute contains index of file references for files and subdirectoriesfor files and subdirectories

NTFS filenameStandard info Index rootSecurity desc. EmptyIndex of files

file1, file2, file3,...

MFT file record for a small directory

If file attribute does not fit into MFT: NTFS allocates separate cluster (run, extent) to store the values NTFS allocates additional runs if an attributes value later grows Those attributes are called non-resident Header of non-resident attribute contains location info

53

Large files & directoriesLarge files & directories

Only attributes that can grow can be non-residentOnly attributes that can grow can be non-resident

Filename & standard info are always residentFilename & standard info are always resident

Index of files for directories forms B+ tree Index of files for directories forms B+ tree

NTFS filenameStandard info HPFS extended attr.Security desc. Data

MFT record for large file with 2 data runs

NTFS filenameStandard info Index rootSecurity desc. Bitmap

file4, file8

MFT file record for a large directorywith nonresident filename index

Index allocation

file1, file2, file3 file5, file6

Index of filesVCN-to-LCN

mappings

54

Large files (contd.)Large files (contd.)NTFS keeps track of runs by means of VCNNTFS keeps track of runs by means of VCN(Virtual Cluster Numbers)(Virtual Cluster Numbers)

Logical Cluster Numbers represent an entire volumeLogical Cluster Numbers represent an entire volumeVirtual Cluster Numbers represent clusters belonging to one fileVirtual Cluster Numbers represent clusters belonging to one fileAttribute lists may extend over multiple runs (not only data)Attribute lists may extend over multiple runs (not only data)

NTFS filenameStandard info Security desc. Data

DataData

VCN-to-LCN mappings for anonresident data attribute

StartinStarting VCNg VCN

StartinStartingg LCNLCN

Number of Number of clustersclusters

00 13551355 44

44 15881588 44

VCN 0 1 2 3

LCN 1355 1356 1357 1358 DataData

VCN 4 5 6 7

LCN 1588 1589 1590 1591

55

Data CompressionData Compression

NTFS supports compressionNTFS supports compression

Per-file, per-directory, per-volume basisPer-file, per-directory, per-volume basis

NTFS compression is performed on user data only, NTFS compression is performed on user data only, not NTFS metadatanot NTFS metadata

Inspect files/volume via Winndows API:Inspect files/volume via Winndows API:GetVolumeInformation(), GetCompressedFileSize()GetVolumeInformation(), GetCompressedFileSize()

Change settings for files/directories:Change settings for files/directories:DeviceIoControl() DeviceIoControl()

with flags with flags FSCTL_GET_COMPRESSION, FSCTL_SET_COMPRESSIONFSCTL_GET_COMPRESSION, FSCTL_SET_COMPRESSION

56

Compression of sparse filesCompression of sparse files

NTFS zeroes all file contents on creation (C2 req.)NTFS zeroes all file contents on creation (C2 req.)Many sparse files contain large amount of zero-bytesMany sparse files contain large amount of zero-bytes

These bytes occupy space on disk unless files are compressedThese bytes occupy space on disk unless files are compressed

NTFS filenameStandard info Security desc. Data

DataData

StartinStartingg VCNVCN

StartinStarting LCNg LCN

Number of Number of clustersclusters

00 13551355 1616

3232 15881588 1616

4848 9696 1616

128128 324324 1616

VCN 0 1 2 3 .... 15

LCN 1355 1356 1357 1358 .... 1370

DataDataVCN 32 33 34 35 ... 47

LCN 1588 1589 1590 1591 .... 1603Certain ranges of VCNs have noCertain ranges of VCNs have nodisk allocation (16-31, 64-127)disk allocation (16-31, 64-127)

57

Compressing Nonsparse DataCompressing Nonsparse DataNTFS divides the files unprocessed data intoNTFS divides the files unprocessed data intocompression unitscompression units 16 clusters long 16 clusters longCertain sequence might not compress muchCertain sequence might not compress much

NTFS determines for each compression unit whether it will shrink NTFS determines for each compression unit whether it will shrink by at least on clusterby at least on clusterIf data does not compress, NTFS allocates cluster space and simply If data does not compress, NTFS allocates cluster space and simply writes datawrites dataIf data compresses at least one cluster, NTFS allocates only the If data compresses at least one cluster, NTFS allocates only the clusters needed for compressed dataclusters needed for compressed data

When writing data, NTFS ensures that each run begins on virtual 16-When writing data, NTFS ensures that each run begins on virtual 16-cluster boundarycluster boundary

NTFS reads/writes at least one compression unit when accessing a NTFS reads/writes at least one compression unit when accessing a filefileRead-ahead + asynch. decompression improves performanceRead-ahead + asynch. decompression improves performance

58

Data runs of a compressed fileData runs of a compressed file0 15

19 20 21 22

VCN

LCN

Compressed data

16 31

23 24 25 26

Compressed data

32 47

97 98 99 100 101 102

Noncompressed data

48 63

Compressed data

27 28 29 30

103 104 105 106 107 108 109 110 111 112

113 114 115 116 117 118 119 120 121 122

StartinStarting VCNg VCN

StartinStarting LCNg LCN

No. of No. of clusterclusterss

00 1919 44

1616 2323 88

3232 9797 1616

4848 113113 1010

MFT record for a compressed file

59

Windows - NTFS ExtensionsWindows - NTFS Extensions

Disk quotas on per-user basesDisk quotas on per-user basesSecurity descriptors (ACLs) can be stored once but referenced in Security descriptors (ACLs) can be stored once but referenced in multiple filesmultiple filesNative support for properties (OLE), including indexingNative support for properties (OLE), including indexingReparse points implementation of symbolic linksReparse points implementation of symbolic links

Mount points for arbitrary file system volumesMount points for arbitrary file system volumesSupport for sparse filesSupport for sparse filesDistributed link tracking (via global object Ids)Distributed link tracking (via global object Ids)

Renaming the target file will no longer break links Renaming the target file will no longer break links (shortcuts...)(shortcuts...)

Add disk space to an NTFS volume without rebootAdd disk space to an NTFS volume without rebootNo decompressing when transmitting files over networkNo decompressing when transmitting files over network

60

File System Driver ArchitectureFile System Driver Architecture

Local File System Drivers (Local FSDs):Local File System Drivers (Local FSDs):Ntfs.sys, Fastfat.sys, Udfs,sys, Cdfs,sysNtfs.sys, Fastfat.sys, Udfs,sys, Cdfs,sysResponsible for registering with the I/O manager and volume Responsible for registering with the I/O manager and volume recognition/integrity checksrecognition/integrity checksFSD creates device objects for each mounted file system formatFSD creates device objects for each mounted file system formatI/O manager makes connection between volumes device objects I/O manager makes connection between volumes device objects (Created by storage device) and the FSDs device object(Created by storage device) and the FSDs device objectLocal FSDs use cache manager to improve file access performanceLocal FSDs use cache manager to improve file access performanceDismount operation permits the system to disconnect FSD from Dismount operation permits the system to disconnect FSD from volume object volume object

When media is changed or when application requires raw device When media is changed or when application requires raw device accessaccessI/O manager reinitiated volume mount operation on next access I/O manager reinitiated volume mount operation on next access to mediato media

61

Layered Drivers - Layered Drivers - I/O System ArchitectureI/O System Architecture

Environmentsubsystem or

DLL

Services

I/O manager

1)1)Call I/O serviceCall I/O service

2)The I/O manager creates an IRP, 2)The I/O manager creates an IRP, initializes first stack location and initializes first stack location and calls file system drivercalls file system driver

3)File system driver fills in a 23)File system driver fills in a 2ndnd IRP stack location and callsIRP stack location and callsthe disk driverthe disk driver

User modeUser modeKernel modeKernel mode

IRP

File systemdriver

Diskdriver

IRP

4)Send IRP data to device4)Send IRP data to device(or queue IRP), and return(or queue IRP), and return

6)Return I/O pending status6)Return I/O pending status

5)Return I/O pending status5)Return I/O pending status

7)Return I/O pending status7)Return I/O pending status

Optimization: associated IRPsOptimization: associated IRPsmay work in parallel on a singlemay work in parallel on a singleI/O requestI/O request

62

File System Driver Architecture File System Driver Architecture (contd.)(contd.)

Remote File System Drivers (Remote FSDs):Remote File System Drivers (Remote FSDs):Client-side FSD translates I/O requests from applications Client-side FSD translates I/O requests from applications into network file system protocol commandsinto network file system protocol commandsServer-side FSD listens for network commands and Server-side FSD listens for network commands and issues I/O requests to local FSDissues I/O requests to local FSDWindows client-side remote FSD: LANMan RedirectorWindows client-side remote FSD: LANMan Redirector

Implemented as port/miniport driver Implemented as port/miniport driver Includes Windows service WorkstationIncludes Windows service Workstation

Server-side FSD server: LANMan ServerServer-side FSD server: LANMan ServerIncludes Windows service ServerIncludes Windows service ServerCIFS common internet file system (enhancement CIFS common internet file system (enhancement of Server Message Block protocol)of Server Message Block protocol)

Application

I/O manager

Remote FSD(redirector)

Local FSD

Remote FSD(server)

Storage devicedrivervolume

user mode

kernel mode

client server

63

Windows Remote File Drivers: Windows Remote File Drivers: Server Message Block (SMB) protocolServer Message Block (SMB) protocol

SMB is a client server, SMB is a client server, request-response protocol.request-response protocol.

The only exception to the request-response nature of SMB The only exception to the request-response nature of SMB is when the client has requested opportunistic locks (oplocks) and is when the client has requested opportunistic locks (oplocks) and the server subsequently has to break an already granted oplock the server subsequently has to break an already granted oplock because another client has requested a file open with a mode that because another client has requested a file open with a mode that is incompatible with the granted oplock. is incompatible with the granted oplock.

In this case, the server sends an unsolicited message to the client In this case, the server sends an unsolicited message to the client signaling the oplock break.signaling the oplock break.

Addl. info athttp://anu.samba.org/cifs/docs/what-is-smb.html

64

SMB and the OSI modelSMB and the OSI model

Clients connect to servers using TCP/IP (actually NetBIOS over Clients connect to servers using TCP/IP (actually NetBIOS over TCP/IP as specified in RFC1001 and RFC1002), NetBEUI or TCP/IP as specified in RFC1001 and RFC1002), NetBEUI or IPX/SPX. IPX/SPX. SMB was also sent over the DECnet protocol. SMB was also sent over the DECnet protocol. Digital (now HP) did this for their PATHWORKS productDigital (now HP) did this for their PATHWORKS product

65

SMB characteristicsSMB characteristics

NetBIOS NamesNetBIOS NamesIf SMB is used over TCP/IP, DECnet or NetBEUI, then NetBIOS names must If SMB is used over TCP/IP, DECnet or NetBEUI, then NetBIOS names must be used in a number of cases. be used in a number of cases. NetBIOS names are up to 15 characters long, and are usually the name of the NetBIOS names are up to 15 characters long, and are usually the name of the computer that is running NetBIOS. computer that is running NetBIOS. Microsoft, and some other implementers, insist that NetBIOS names be in Microsoft, and some other implementers, insist that NetBIOS names be in upper case, especially when presented to servers as the CALLED NAME. upper case, especially when presented to servers as the CALLED NAME.

Protocol functionality (Core protocol):Protocol functionality (Core protocol):connecting to and disconnecting from file and print shares connecting to and disconnecting from file and print shares opening and closing files opening and closing files opening and closing print files opening and closing print files reading and writing files reading and writing files creating and deleting files and directories creating and deleting files and directories searching directories searching directories getting and setting file attributes getting and setting file attributes Locking and unlocking byte ranges in files Locking and unlocking byte ranges in files

66

SMB characteristics (contd.)SMB characteristics (contd.)

SecuritySecurityThe SMB model defines two levels of security: The SMB model defines two levels of security: Share level. Share level.

Each share can have a password, and a client only needs that password Each share can have a password, and a client only needs that password to access all files under that share. to access all files under that share. This was the first security model that SMB had and is the only security This was the first security model that SMB had and is the only security model available in the Core and CorePlus protocols. model available in the Core and CorePlus protocols.

User Level.User Level.Protection is applied to individual files in each share and is based on Protection is applied to individual files in each share and is based on user access rights. user access rights. Each user (client) must log in to the server and be authenticated by the Each user (client) must log in to the server and be authenticated by the server.server. When it is authenticated, the client is given a UID which it must present When it is authenticated, the client is given a UID which it must present on all subsequent accesses to the server. on all subsequent accesses to the server. This model has been available since LAN Manager 1.0. This model has been available since LAN Manager 1.0.

67

SMB Clients and Servers SMB Clients and Servers

Clients:Clients:Included in WfW 3.x, Win 95, Win98, Win ME and Windows Included in WfW 3.x, Win 95, Win98, Win ME and Windows NT/2000/XP/Server 2003/Vista. NT/2000/XP/Server 2003/Vista. smbclient from Samba, smbfs for Linux, SMBlib smbclient from Samba, smbfs for Linux, SMBlib

Servers:Servers:Microsoft Windows for Workgroups 3.x, Win95, Win98, Win ME, Microsoft Windows for Workgroups 3.x, Win95, Win98, Win ME, Windows Windows NT/2000/XP/Server 2003/VistaNT/2000/XP/Server 2003/Vista SambaSamba (Linux, Solaris, SunOS, HP-UX, ULTRIX, DEC OSF/1, Digital (Linux, Solaris, SunOS, HP-UX, ULTRIX, DEC OSF/1, Digital UNIX, Dynix (Sequent), IRIX (SGI), SCO Open Server, DG-UX, UNIX, Dynix (Sequent), IRIX (SGI), SCO Open Server, DG-UX, UNIXWARE, AIX, BSDI, NetBSD, NEXTSTEP, A/UX)UNIXWARE, AIX, BSDI, NetBSD, NEXTSTEP, A/UX)The PATHWORKS family of servers from Digital The PATHWORKS family of servers from Digital LAN Manager for OS/2, SCO, etc LAN Manager for OS/2, SCO, etc VisionFS from SCO VisionFS from SCO Advanced Server for UNIX from AT&T (NCR?) Advanced Server for UNIX from AT&T (NCR?) LAN Server for OS/2 from IBM LAN Server for OS/2 from IBM

68

Further ReadingFurther Reading

Mark E. Russinovich and David A. Solomon, Mark E. Russinovich and David A. Solomon, Microsoft Windows Internals, 4th Edition, Microsoft Microsoft Windows Internals, 4th Edition, Microsoft Press, 2004.Press, 2004.

File Systems supported by Windows (from pp. 689)File Systems supported by Windows (from pp. 689)File System Driver Architecture (from pp. 694)File System Driver Architecture (from pp. 694)

NTFS Design Goals and FeaturesNTFS Design Goals and Features (from pp. 717)(from pp. 717)NTFS On-Disk Structure (from pp. 732)NTFS On-Disk Structure (from pp. 732)

69

Source Code ReferencesSource Code References

Windows Research Kernel sources do not Windows Research Kernel sources do not include NTFSinclude NTFS

A raw file system driver is included in A raw file system driver is included in \base\ntos\raw\base\ntos\raw

Also see \base\ntos\fstrl (File System Run-Time Also see \base\ntos\fstrl (File System Run-Time Library)Library)

Unit OS8: File SystemCopyright Notice 2000-2005 David A. Solomon and Mark RussinovichRoadmap for Section 8.2Windows File System - TerminologyFormats Supported by WindowsCDFSUDF FAT FAT12FAT16 FAT32NTFSCIFS the Common Internet File SystemFile System Format CompatibilityNTFS Design GoalsNTFS RecoverabilityNTFS Security and RecoverabilityLarge Disks and Large FilesLarge Disks and Large Files (contd.)Other NTFS FeaturesMultiple Data StreamsSlide 22Unicode NamesHard LinksJunctionsSlide 26Change LoggingSlide 28Per-User Volume QuotasLink TrackingEncryptionPOSIX SupportDefragmentationSlide 34Compression and Sparse FilesNTFS File System DriverComponents related to NTFSNTFS & File ObjectsNTFS On-Disk StructureNTFS Cluster SizeMaster File TableNTFS operationNTFS metadataNTFS metadata (contd.)File Records & File Reference NumbersFile Records (contd.)Standard Attributes for NTFS FilesAttributes (contd.)FilenamesMS-DOS filenames in NTFSResident & Nonresident AttributesSlide 52Large files & directoriesLarge files (contd.)Data CompressionCompression of sparse filesCompressing Nonsparse DataData runs of a compressed fileWindows - NTFS ExtensionsFile System Driver ArchitectureLayered Drivers - I/O System ArchitectureFile System Driver Architecture (contd.)Windows Remote File Drivers: Server Message Block (SMB) protocolSMB and the OSI modelSMB characteristicsSMB characteristics (contd.)SMB Clients and Servers Further ReadingSource Code References