Raid data recovery

RAID Data Recovery

Most motherboards incorporate RAID controllers, but they are still little used. This technology allows, however, for a reasonable cost, to ensure the security of data and increase performance of the computer.

Generally, only one disk is used for the operations of writing and reading but technology RAID (Redundant Array of Independent / Inexpensive Disks) uses several mounted cluster. There are different types of RAID technology that will be used depending on whether we wish to emphasize performance, data security, or why not both. Given the low cost of hard disks, these techniques are inexpensive. The introduction of a RAID array is very simple. Simply launch the BIOS, forming RAID that we have chosen and to indicate the kind of discs to be fitted. Next, the system is totally transparent to the user, RAID array being viewed simply as a single disk. It should be noted that the construction of a cluster generally destroys all data on disk (except sometimes for RAID 1) and that the discs must be of the same capacity, and possibly identical model to avoid wasteful important.

Speed or safety: we must choose!

The first is the technology used again called RAID 0 striping. It involves cutting a file into small strips (strips) and how to write them alternately on two or more hard disks. For example, during a drive writes data, another piece is sent to another disk, and so forth. Each disk had to write a portion of the data file, access is much faster in both reading and writing. The more disks, the greater the performance is to the appointment, but in principle two or three hard enough. RAID 0 allows high performance, even with ordinary discs, much less expensive than high-end drives to 10000 rpm. However, the system suffers from a lack of reliability. If a disk fails, all of the information in the cluster is lost. With three disks, the risks of failure are three times greater than just one. RAID 0 therefore should be used only on volumes where data loss is not catastrophic as the operating system or programs. With identical drives, the ability of the cluster is equal to the sum of the capacities of the disks. But if one goes up in RAID 0 disk 160 GB and 250 GB would yield a total capacity of only 320 GB The 90 GB most of the largest disc sit unused.

If data security prevails, it is necessary to go through the mirroring or RAID 1. In this mode, the controller writes the same data on multiple disks. If one of them fails, the data is fully recoverable on the remaining disks. Simply replacing the hard disk corrupted by a nine per controller that regenerates over the system. Two discs in principle sufficient to ensure adequate security. Another advantage of RAID 1 is in the case of failure the user can continue to work while awaiting the installation of a new disc and even while the system regenerates. In contrast, the RAID 1 is not very economical, since the total capacity equals that of the smallest disks in the cluster. In our example, with two drives of 160 GB and 250 GB, the total capacity will be only 160 GB



Performance and Safety: you can't have your cake and eat it

How can I get both the performance and safety? Just mount first two sets of disks in RAID 1 for the safe and put everything in RAID 0 to gain speed. The result is a series called RAID 10 or RAID 1 +0. It is easier to regenerate that the opposite is to RAID 1 cluster RAID 0 (mode 0 +1). The 10 combines RAID performance and security, but it needs a minimum of four drives, which is not very economical. With identical drives, the total capacity is twice that of a disk, or 320 GB for example when using four drives of 160 GB

There is another way to combine performance and security: the RAID 5. The principle is a bit more complex and requires a minimum of three disks. As with RAID 0, the data are cut into strips. The first strip is saved on a disk, a second on another disk and a third disc is a parity computed on the basis of an "exclusive or" between the first two strips, which is recorded. Registration continues into rotational disks, checks are distributed equally on all disks. If one of them fails, the missing information will be replenished by performing the same operations "exclusive or" between the strips to replenish the controls, or between controls and strips to reconstruct the missing strips.

What to do in case of breakdown?

In case of failure, we must first replace the damaged disk before resuming his work. Secondly, it should be pointed out that the recovery of the data is fairly long, it usually lasts several hours. RAID 5 is interesting, because it is cheaper than RAID 10, but a little more difficult to regenerate in the event of failure. With identical drives, the total capacity is equal to the sum of that of at least one disk. So with three disks of 160 GB leads to a total of 320 GB ((3-1) * 160) and with four 480 GB disks to only 320 in RAID 10. The RAID clusters consist of whole disks, but thanks to the Intel Matrix Storage Technology, it is possible to mount several different RAID systems on the same group of disks. With three discs for example 160 GB, we can create a RAID 0 on the first 60 gigabytes of each disc is 180 GB and 100 GB place that remain in RAID 5 is 200 GB The Intel Matrix Storage RAID makes even more economical, but it can not be implanted on the motherboard chipset with Intel ICH6R or later.

You might want to read our previous post about Raid Data Recovery

What is data recovery?

Data recovery software has multiple motivations: the most common range from inadvertent human error, viruses or Police investigation. For example, in the Clearstream affair in France, a portion of the survey was based on data retrieved from the computer of General Rondeau.

The principle of data recovery is as follows: in general, operating systems cut all resources into pieces which they access through index. Thus, very roughly, an operating system, sees a hard drive like an encyclopedia, and the system will use the table of contents of the encyclopedia to quickly access a particular article according to the demands of the users. When the user requests the operating system to erase a structure (a file or a partition, for example), the operating system does not destroy the structure directly concerned. It merely changing the index. Thus, in the metaphor of the encyclopedia, ask the operating system to delete an item, the equivalent of removing the item from the table of contents. The article dropped, even if not included in the table of contents, however, is always there in the middle of the pages of the encyclopedia.

Data Recovery is a difficult task. Indeed, in the presence of backups - whether copies stored in the dustbin of Windows or any tool System Restore, it is beyond the scope of data recovery. In addition, operating systems are primarily designed to use resources effectively to ensure good performance. For example, the user can request the free space to the operating system (to add a new article in the encyclopedia, for example). In this case, the operating system can give pages corresponding to the item deleted the user wants to write a new article. As operating systems are designed primarily to ensure good performance, data recovery is at best an afterthought. It is significant that from version 5, the late MS-DOS provides two tools for data recovery ( "undelete" and "unformat"). Both tools were included in version 6 but were abandoned to the launch of Microsoft Windows 95.