Knowledge, Filesystems

1833 words [ 2 Weblinks ] - Last update: 2016-10-16 Page created: 2004-04-03 [SB]

File systems

File systems are needed to store files and directories organized to any data storage media. Read and write accesses on the file system are realized by the driver software. Worldwide exists more than 100 different file systems and partition types with different features. Additionally are network file systems which exist only within net environments and make the data access independent from the platform on protocol layer possible. The file systems for parallel computing also belong to the network field. Examples of such are GPFS (General Parallel File System) of IBM for the operating system AIX, PVFS (Parallel Virtual File System) for Linux Cluster or also the GFS (Global File System) to name only a few.

BFS | ext2 | FAT 12,16,32 | HFS | HPFS | ReiserFS | NFS | NTFS
fs-QNX | iFS | NFS | SMB |

BFS Be file system The Be file system (BFS) is a 64-Bit file system which was also like BeOS designed for the multithread ability. With the journaling function all file system accesses are stored in a kind of database which offers high performance for the access to attributes of files and directories. If a error occured, inconsistencies can be avoided or cleared. Even incomplete data storage processes can be undone. A great unusual feature is the file type identification. It is not fixed on the file extension but about the MIME attribute which is used also for the marking of e-mails. About the definition of the MIME type with an application the file can be opened in BeOS. If the file type is not assigned, BeOS fetches the file extension and signed it to an MIME type.
- Partition size max. 2 64 bytes (approx. 18 trillions gbyte)

ext2-fs Extended filesystem ext2 is the extended file system of Rémy Card. It can be described as the standard file system for the Linux operating system. The file names can contain up to 255 characters, the maximum size for a partition with kernel 2.2 are at 2 gbyte.
This file system is organized that to each file exists an information file (inode) which contains evertything except for the filename. So the access rights, file size, position and the occupied file blocks are notized.
In meanwhile the 3th version of the ext file system is used with many improvements in the area of design and performance.

FAT 12/16/32 File Allocation Table FAT file systems are built up with a simply structure and include an additional copy of the file allocation table in the first mbyte on the data storage media. At damage to the main table this can be corrected by the second copy, on condition that the first mbyte starting at sector 0 was not deleted. The access to FAT partitions is supported by most operating systems as a standard. This file system uses only very low resources, only the high slack space with to big cluster of formated storage devices is a disadvantage. File names are generally stored in capital letters, a differentiation distinction of small and capital letters is not made under Windows operating systems. FAT file systems supports only few attributes. These are R, H, S, A: Read-only, Hidden, System and Archive.

The number of at most addressable Clusters in bit with the FAT file system is described by the 2-digit number. From this the following details are shown:

FeaturesFAT 12FAT 16FAT 32
max. file size:32 mbyte2,048 mbyte4,096 mbyte
max. partition size:32 mbyte2,048 mbyte (DOS)
4,096 mbyte (NT)
8,192 gbyte
max. number of files:409665,536approx. 4.2 mio.

FAT 12 Was used for the first time for 180 kbytes floppy disks in 1976. The idea and basics for this file system origin from Tim Patterson, the technique was bought up 1976 of Bill Gates. For the file name at most 8 characters and 3 characters for the extension are possible.

FAT 32 is the successor of FAT 12 for hard disks. This revised version allowes partition sizes of up to 8 terabytes now. Formatting more than 32 gbyte big FAT32 partitions is not possible in Windows 2000 and Windows XP directly. Opposite to FAT 16 FAT 32 uses smaller clusters which leads to a better use of storage and less slack space. By the a little risen administration overhead it is usable with at least 90 mhz at the approximately same performance at the same combatibiliy. File names are not restricted to the 8.3 scheme any more.

HFS Hierarchical Filesystem HFS is the standard file system for Macintosh computer. It is the successor of the MFS (Macintosh file system) file system. A partition can be formatted with up to 65,535 Cluster (16 bit), the cluster size varies according to the size of the partition. File names can consist of up to 31 characters, the maximum file size is 2 gbyte. The successor is called HFS+ and records all accesses to the file system in a journal for a higher data security. The file name can consist of up to 255 unicode character now. The file system can address clusters with 32-Bit and supports a maximum file size of 8 exabyte (263 byte).

HPFS High Performance File System This file system was developed together by IBM and Microsoft around the year 1985. At first it was used in OS/2 1.2 and Microsoft LAN Server. The idea arose from the insufficient abilities of the FAT file system for use on servers and networks, like missing file access rights. HPFS has a considerably more progressive method to manage files and directories and avoid the fragmentation largely. By the order in B-trees objects can be located much faster. The object table is in addition also located in the middle of the partition to increase the data security and minimise the seek time. Own attributes like informations about the origin of a file can be defined. In 1991 the cooperation on OS/2 by Microsoft was quit which led between IBM and Microsoft to a break. New sales partner for OS/2 since then was Apple.

ReiserFS ReiserFS belongs to the journaling file systems. With his administration structure built up like a database any logged change can be undone. All accesses to the file system are included and provides moreover the consistency of data also at a sudden power failure. The check on file consistency is very fast. Linux, HP-UX, AIX and OS/2 used this file system it optional and take advance of the high performance, reliability and fast access time speed at the storage of large amounts of data. It cooperates also reliably with the different RAID levels in the software mode.

NTFS New Technology File System This file system is used since the first Windows NT version. It has design concepts of HPFS inherited. The administration takes place directly at sectors area and makes a better storage usage possible. To the administration of the sectors and files the MFT (Master File Table) comes to use. It belongs to the journaling file systems which take down and supervise every modification. If a process is not finished completely, the original status is restored. With NTFS single files and folders can be signed with rights. The partition size can be up to 2 64 byte, file names with up to 255 unicode characters are possible. A transparent compression of files and directories is supported and offers a transaction administration of every access by indizies. Attributes of stored objects are Read, Write, Execute, Delete, Owner, Permission and Compressed.

NTFS features (since Windows XP)
- Revised functions of the pre-version
- Encryption of the file system with EFS
- Data storage quotas (taken by UNIX, disk quotas)
- Provision points (taken by UNIX, mount points)
- Supervision of the path of distributed links
- Transparent compression of files with a low data density
- Modification journal for file operations

fs-QNX QNX Filesystem This file system is very efficient and has an extremely robust design. The file allocation table is a bitmap which prevents data losses and simplifies a recovery by signatures (control structures). It is like the QNX 4 file system, contains POSIX features and supports multi-threading. File names can consist of up to 48 characters.

iFS Internet File System The Oracle iFS stores files in an Oracle 9i database, but can manage also external files and include it in the general use and administration of files in the network as well as the extensive search function. The access is also possible over different protocols like HTTP, SMB, FTP and SMTP. All functions can be used only over the special Oracle software. Additional features like management of different file versions and XML support counts to the strengths of this file system. The customization ability and conversion of documents as well as e-mails are also noteworthy. The read and write of files within the database is generally slower than at pure file servers. In return the search function leads considerably faster at great archives to a result.

NFS Network Filesystem Sun Microsystems developed this distributed file system protocol to make the file access over several computers possible. With this network file system directory entries of different computers can be exchanged in the local network. It is the standard file system of computers in UNIX networks for the data access.

SMB Server Message Block This network protocol make it possible to access the exported directories of a computer with NetBIOS and TCP/IP support, it is indirect also an file system. Windows for example offers this functionality as of the version 3.11 (Workgroups) if a TCP/IP-stack is installed. Homogeneous Windows networks but also heterogeneous networks which consist of UNIX or related operating systems and Windows computers are main fields of use.


[ XFS ][ ReiserFS ]