File NTFS system. Principles of construction.

Before Windows NT 3.1 released, Microsoft invented two file systems: FAT (File Allocation Table) based on MS-DOS and Windows, and High Performance File System (HPFS) used in OS/2 operating system. As Windows NT coming to the market, Microsoft needed a new file system to support the security and reliability of NT. FAT and HPFS obviously were born to have defects in this. After deliberation, the design team decided to create a brand-new file system with good tolerance and security - NTFS (New Technology File System). From Windows NT 3.1 to Windows 2000/XP, NTFS was being developed unceasingly. Compared with FAT and HPFS, besides the enviable performance features and backward compatibility, NTFS also becomes the first file system that provides sound file service for high server and Intel workstation family.

From the very beginning, NTFS is set for enterprise file system. In order to reduce data loss by sudden power cut or system collapsing, the file system should always guarantee the integrity of metadata in file system. In order to protect the sensitive data from illegal visit, there should be a comprehensive security model in file system; In order to protect the user data, the file system should provide inexpensive data redundancy plan based on software instead of expensive one based on hardware.

High-level features of NTFS

1. Multi-data streams
2. Name based on Unicode
3. General index mechanism
4. The dynamic bad cluster reprints maps
5. Supports POSIX
6. File compression
7. File encrypts
8. Disk quota
9. Hard link and soft link
10. Link tracks
11. Log records
12. Fragmentation

In NTFS file system; the files are also assigned according to clusters. A cluster must be the integral multiple of physical sector, moreover always integer powers of 2. NTFS file system does not care for about sector, nor sector size (for example not 512 bytes), but cluster size is assigned automatically according to the volume size by format program when formatting.

By master file table (MFT) to determine the file storage location in disk. The master file table is a corresponding database composed of series of file records-each file has a record in a volume (large files may have many corresponding records). The first file record is called basic file record with some information of other extended file records. Master file table itself also has its own file record.

Each file in NTFS volume has a unique identifier of 64 bit called file quotation number/File Reference Number. The file quotation number consists of two parts: file number and file order. The file number is 48 bit, corresponding to the position of MFT. The file order increases as repeating usage of file records, which is designed for internal consistency examination of NTFS.

NTFS locates clusters with logic cluster number (LCN) and virtual cluster number (VCN). LCN is the simple number for all clusters from beginning to end. With volume factors multiplying LCN, we can get physical byte offset in volume, thus obtaining physics disk address. VCN is the number for specified files from beginning to end, which is convenient for quoting the data of files. VCN can be mapped as LCN, but it does not request physical continuity.

Directory of NTFS is only a simple filename and index of the file quotation number. If the directory attribute list is smaller than the length of a record, then all information of this directory are saved in the main file table records, as to that bigger than record directory, it was managed with B+ tree.

In the basic file recording of master file table, it has an indicator aiming at the index cache which is not usually used, including sub-directory and file exterior cluster, but B+ tree structure is convenient for the file and the sub-directory in large-scale table of contents.

In NTFS, all data saved in the volume are contained in file, including file construction that is used to locate and obtain files, boot program and bitmap files that record volume size and service condition. This manifests the principle of NTFS: all things in disk are called files. Saving everything in files makes the file system very easy to locate and maintain data. The file fixes its storage location in disk with master file table.

The relationship of NTFS' each region

Size of file records in MFT is generally fixed. No matter the size of a cluster, the records all are 1KB, which equals to inode (i node) in Linux. File records in MFT file record array are physically continual, number from 0. Therefore NTFS can be considered as the pre-definition file system. MFT is only used by system's organization and framework file system. This is called Metadata in NTFS.

Date: 2015-01-29; view: 989