Recovery UNIX - Linux - 2

File Systems

EXT 2 is the historical system of the Linux platform and was marketed in 1993. It has since been replaced by E 3, which has added some new features including the most notable being the logging. EXT3 is the default file system for most Linux distributions.

Like many other UNIX file systems, the main structure is very similar to the original UNIX Fast File System (FFS). The score is divided into groups of cylinders, and each of these groups contains initially Superblock, Group Table, the data bitmap, a bitmap node, node files and data. However, more recent versions of E 2 and Group 3 have hollow cylinder that contains only nodes and data files.



The EXT file systems 2 and 3 have a fixed number of nodes file and they are drawn on the score by the superblocks and Group Tables. These nodes files are data structures containing information about nt files stored in some file systems:

* The file type
* Access rights
* Owners
* The timestamps
* The size
* The data block pointers



The pointer data block are the node which sends the data file on disk.


The first twelve are pointing to the physical blocks of data. The first 12 fields (out of 15) contain the addresses of the first 10 blocks of file data (with an address block). If the blocks on which point the first 10 fields are sufficient to contain the file, the fields 11, 12 and 13 are not used. H owever, the three point to indirect blocks of data (single, double and triple indirect ion s). L a simple indirect ion contains the address of a block of direct pointers (called block address) as shown in the diagram. Thus, the address block contains pointers to data blocks (256 pointers). The double indirect ion points to a single block containing single indirect pointers. However, this block of addresses is pointing to 256 blocks of data it points to other blocks of 256 addresses. These are the 256 blocks of addresses that point to 256 blocks of data. L ogiquement triple indirect ion points to a block of double indirect pointers (to follow the same reasoning as above). This can be very difficult to visualize, but basically every step of indirection allows the amount of data to be sent to the exponential.

XFS

XFS was originally developed by SGI in 1993. The project was to overcome some limitations in terms of performance and extension of the FFS. This is a file system journaling 64-bit high performance. It was first published in 1994 with IRIX V5.3 and 2000, SGI released the source code open. It was then formally included in the Linux kernel since 2003. The structure of XFS is very similar to that of FFS at first sight. It keeps the system of cylinders Groups splitting the partition, but the groups named for the award and he also uses superblocks and filesystem nodes to contain the metadata files, but the similarities in content and stop functioning at that level .

Unlike FFS, the file system does not have a fixed number of knots of pre-allocated files on disk. In its place, each group should monitor the allocation space and dynamically allocate nodes file according to the needs expressed by the file system. These nodes files are organized in a balanced tree B +, which makes traversing the directory structure much faster than the traditional list system implemented in FFS. However, to maintain high performance, the B + tree is kept balanced thanks to a grant of intelligent nodes and files it requires relatively advanced algorithm. The nodes also use XFS file extension (run lists) to send data instead of treating the individual blocks of data as FFS, and it adjusts to the best size for your important files.

XFS journaling also provide an opportunity for recovery of file system in case of crashes and power outages. However XFS reviewed only the file system metadata while the volume can be repaired and loaded on it and can still be a cause of data loss.

Another feature of XFS has delayed allocation. First, the data are stored in memory. Then these data is actually written to the file system where the cache is cleared by the operating system. This method of allocation wave allows for a certain delay in the action to obtain the necessary data to allocate the most intelligent way possible. The main advantages of this approach is the dramatic decrease of fragmentation, especially with files that grow slowly and often reduces the CPU overhead.

JFS (Journaled File System)


In 1990, IBM produced a first draft JFS with AIX version 3.1. Later in 1999, IBM brought it under OS / 2 and published a version of JFS in the open source community and in 2006 was the stable version of Linux.


The philosophy behind the design JFS is comparable to that of XFS. They aim to overcome many limitations of FFS performance in a very similar way, although recent implementations are different. Both file systems use metadata logged to provide the file system is a possibility of recovery, nodes dynamically allocated files and extensions to address the area of data and also the diagram of the transverse B + tree. For more information, see the section XFS.

LVM

Managing logical volumes (in English, logical volume management or LVM) is a method to overcome some limitations of using traditional methods partition to allocate storage space on the media. Generally the features included are:

* File System Spanning and software RAID (Level 0, 1 and 5)
* Resize volume groups and logical volumes
* Photo (the hard disk and partition are concerned) or snapshots

Traditionally space on hard disks is divided into partitions on which file systems are written directly. LVM works a little differently, the disks are always allocated using them, but they are considered "physical volumes" on LVM. Hard drives, hard disk partitions, RAID volumes or logical units from a SAN form the physical volumes. Then we concatenate these physical volumes in "volume groups". The volume group can then be allocated to a form of logical volumes on which file systems reside. The diagram below shows a relatively simple example of how LVM could be used.



Nowadays, cases of UNIX versions of adopting their own variations of LVM are very common. According to the seller they have names and different functions. Linux also has an LVM which is based on the Hewlett Packard UNIX version. One feature missing on both the HP and the Linux LVM is that they have no implementation to manage fault tolerance by gender, so no software RAID 5.

Windows 2000, 2003, XP and Vista have an equivalent system called Logical Disk Manager, which offers similar capabilities.