What about file systems-2nd unique file system features

Determining whether a file system has certain characteristics is an effective way to distinguish and Compare file systems from different aspects. Some common file system features are as follows.
 
2.1 hard link
A hard link is a reference or pointer to a file on a storage volume ). The name associated with the file is a tag stored in the directory structure. This label points the operation to the data of the file. In this way, the same file can be associated with multiple names. When these different names are used, any changes made will affect the data of the same file. There are many restrictions on using hard links. First, to prevent endless recursion, most modern operating systems do not allow hard links to folders. Secondly, the number of hard links is also limited by the number of reference counts. In addition, hard links can only be used on the same file system as files. If you need to link files to another file system, you may need to use soft links.
 
2.2 Symbolic Link
A symbolic link (symbolic link, or symlink, soft link, or soft link) is a special type of file. The file is in the form of an absolute or relative path, point to another file or directory and affect pathname resolution ). The symbolic link differs from the hardware connection in that it is a file and can exist independently of the target file. If you delete a symbolic link file, the target file will not be affected. If the target file is moved, renamed, or deleted, all symbolic links pointing to the target file will point to a non-existent file. This symbolic link is called orphaned or dangling ).
 
2.3 Log File System
Journaling file system is a file system that records modifications to a log before applying modifications to the main file system. This file system is less likely to be damaged in the case of power failure and system crash.
 
Although the log system increases reliability, because the log requires that all data be written twice, the file system performance may be seriously affected. Only the changes in the file metadata are recorded in the log, that is, metadata-only journaling, which is a compromise between reliability and performance. This method can still ensure that the file system can be quickly restored at the next installation. However, due to the absence of file data changes, data corruption may occur or the recorded metadata is not synchronized.
 
For example, adding content to a Unix file system involves the following three steps:
 
1. Increase the file size on the file index node (inode.
 
2. Allocate space for the added content on the free space bitmap.
 
3. Write the additional content into the allocated space.
 
In a metadata-only journaling, step 1 may not be recorded in the log. If Step 1 is not executed, but steps 2 and 3 are redone during restoration, the file will be appended with junk information.
 
2.4 log structure File System
In log-structured file system, all updates to data and metadata are written to a continuous log sequentially. The log structure file system is based on the assumption that the increasing memory capacity in modern computers makes write operations costly because most read operations can be obtained through memory cache at high speed. Therefore, the log structure file system regards the storage as a ring log and writes it in sequence in the log header, this increases the write throughput by avoiding time-consuming seek. To prevent the file system from being fully occupied, free space needs to be recycled from the end of the log.
 
The log structure file system generates constantly updated file data and metadata for multiple versions. The implementation of some log structure file systems allows these versions to be accessed. This feature is called time-travel (snapshoel ). This is very similar to the version file system. The log structure file system simplifies recovery from crashes.
 
The log structure file system also has its disadvantages. Especially when the application is in the disk medium, the tracing overhead is relatively large, which may slow down the reading speed because it separates the files stored continuously in the conventional file system into many fragments.
 
Log structure file system. DF, LFS, NILFS, and other file systems are log structure file systems.
 
2.5 dynamically allocate logs
Dynamically allocates logs (dynamically-allocated journal ). In the log structure file system, the performance loss caused by two writes is not required, because the log itself is a file system. The file structure of most Unix systems is not a log structure, but some file systems use similar mechanisms to avoid performance loss of two writes. For example, Reiser4 can combine many discrete writes into a single continuous mass write, and then expand the log header to attach the newly written mass, the tail of the log is scaled back after the application is applied to the storage.
 
2.6 case sensitive
Case-sensitive ). If the file system is case sensitive, the file name must be given in strict accordance with the case of the original file name and letter when reading the file in this file system. If the file system is case-insensitive, the letters of the given file name can be case-insensitive.
 
2.7 case-sensitive
Case-preserving ). Case sensitivity indicates that the file system stores the name of a file strictly in the case of letters. If a file system is not case-sensitive, it must not be case-sensitive. Conversely, if a file system is case-sensitive, it must be case-sensitive.
 
2.8 file change log
The file change log tracks changes to the file system's namespace (files and folders. Depending on the implementation, the file change log records such as creation, Link, unlink (delete), rename file, data modification, and metadata modification (such as ACL (Access Control List, accesscontrol list) and access permission modification ). Note that the file change log and the file system log are different. By using the file change log, the application does not need to repeatedly scan the entire file system to detect changes in the file system. File Change logs can also be used as audit records for file system activities.
 
2.9 snapshot www.2cto.com
Snapshot is a copy of some files and folders in the file system at a certain time point in the past.
 
2.10 local execution
Execute in place (XIP) refers to executing a program directly in non-volatile memory without copying the program to RAM. Because local execution requires Random Access to the memory, non-volatile memory generally refers to the NOR Flash, because only the NOR Flash can be easily accessed at random.
 
2.11 file system-level encryption
File System-level encryption (filesystem-level encryption) is usually called file or folderencryption. It is a feature that the file system itself encrypts individual files or folders on disks. Unlike full disk encryption, full disk encryption encrypts the entire partition of the file system in full disk encryption.
 
File System-level encryption has the following advantages:
 
1. Flexible file-based key management. In this way, each file can be encrypted using different encryption keys.
 
2. Separate encrypted file management, that is, you can perform Incremental backup on the separately encrypted file, instead of backing up the entire encrypted volume.
 
3. You can use public key encryption to enforce access control.
 
4. The decryption keys are only stored in the memory, and the files they are unlocked can be kept open.
 
2.12 integrated logical volume Management
Logical volume manager (LVM) is a more flexible space allocation mechanism for large-capacity storage devices than the traditional partitioning mechanism. In particular, logical volumes can concatenate and combine partitions to form large virtual partitions that can be moved and resized when they are in use. Logical volume management is one of many forms of storage virtualization.
 
Of course, the introduction of logical volume management makes the startup process complex and makes disaster recovery difficult, especially when the basic operating system and other necessary tools are on the logical volume. The logical volume is also affected by external fragmentation, resulting in lower I/O performance.
 
External fragmentation refers to the process in which idle storage space is divided into several small blocks as time advances. Internal fragmentation refers to the fact that a bucket is never intended to be used after it is allocated.
 
2.12 restrictions on file systems
The file system has some restrictions, such as the maximum file name length, letters allowed for directory items, the maximum path name length, the maximum file size, and the maximum volume size.
 
The length and size are usually limited by the number of bits in the data structure used. Some file systems have no limit on the maximum length of path names.
 
In addition, the letters allowed for directory items are limited by the structure of directory items on the disk. Some installable file system drivers may also limit the names of file names and folders. Some operating systems may also have restrictions on all file systems used. For example, MS-DOS, Microsoft Windows, and OS/2 do not allow any file system to use "\/:? * "> <|" And NULL characters.
 
2.13 metadata
Metadata (metadata, meta data, or metainformation) is the data about the data. The file system uses metadata to manage files.
 
The metadata of the file system may include: file owner, POSIX file access permission, creation timestamp, recent access/read timestamp, recent change content, latest metadata modification timestamp, recent archive timestamp, and access control List list, ACL), Security/mandatory access control (mandatory access control, MAC) Labels, extended file attributes, backup data, and fork), checksum/error correcting code, ECC ).
 
Mandatory Access Control refers to an access control method that supports its file system to limit the ability of the initiator to access or perform certain operations on the target.
 
Extended file attributes refer to the metadata that the file system allows you to associate files with without the need to be interpreted by the file system. Generally, file attributes are strictly defined by the file system.
2.14 transparent Compression
Some file systems support transparent compression, such as Btrfs, NTFS, Reiser4, JFS, NSS, NWFS, Fossil, ZFS, and other file systems.
 
This article is welcome to reprint, please keep the original blog link http://blog.csdn.net/fsdev/article