RAID 5 stripes at the block level but does not use a single dedicated drive for storing parity. Instead, parity is interspersed within the data, so after each sequence of data blocks there is a block of parity data, but this changes to a different disk then for the next set of blocks.
This could means, for example, that in a 3 disk RAID 5 there are data blocks on drive 1 and 2 followed by a parity block on drive 3. For the next set of blocks the data is on drive 1 and 3 with the parity on drive 2, then data on drive 2 and 3 with parity on drive 1.
RAID 5 is generally fast for smaller reads, so eminently suitable for server systems being shared by large numbers of users created smaller data files or accessing smaller amounts of data each time.
From the view of RAID data recovery, the use of error correction RAID 5 increases the likelihood of success greatly. Whilst RAID systems tend to be rather cut-and-dried about disks with errors, often they will just stop using the disks (so if you have 2 disks developing problems the whole RAID could become inaccessible even though the number of sectors that are actually unreadable is very low), the data recovery process can be much more sophisticated. Decisions regarding which sector to use or whether to re-calculate missing data can be made on a sector by sector basis and often this means that a RAID reporting 3 bad disks will be sent in for data recovery and the result will be a 100% success.
Damage caused by contact between the hard drive read/write heads and the disk platter, and flaws within the recorded data surface, can result in disk sectors becoming unreadable. Most RAID5 configurations will treat this occurrence as a disk failure and resort to running in degraded mode, so ceasing to access the disk with errors and re-creating the data it holds "on-the-fly" in response to read requests. If two or more disks report errors then the RAID5 error correction can no longer cope and so the RAID will cease to function. The key point here for RAID data recovery is that only where there are corresponding failed sectors on the drives that have problems will there be loss of data. The data recovery process will involve the securing of all available data from each of the drives and then a rebuilding process that treats all disks as being still active, but that can make decisions on a sector-by-sector basis as to whether the data from all of the drives can be used or, if not, which data to use as part of the rebuilding process.If there are no corresponding bad sectors then the physical data recovery process will be 100% and so if there are no problems with the file system then the work is almost done. In many instances this form of data recovery technique can be used to rebuild the failed disks which can then replaced within the RAID unit and the recovery work is completed.
Reconfiguration and Rebuilding
RAID rebuilding is a process designed to make a RAID work, but has no consideration for the data. If a RAID is offline and will not start it may be reported that a rebuilding is required. This could be the result of the loss of RAID configuration information from the controller. Rebuilding is not a defined process, the precise manner is down to the RAID vendor. It may be that the disks are read and the parity information recalculated and written back, it could be that the disks are wiped as part of the process. If the disks are erased then all data is lost, if the configuration has changed prior to the rebuilding then there will be significant corruption to the data as the pattern of parity distribution across the disks will have changed. The important factor is how long the process took, if quick then there will not have been time for damage to a significant proportion of the data, if there was a process that lasted for several hours then there might have been. The RAID data recovery process entail the securing of all data from the disks and then an examination to determine what has happened and how it has affected the data. A full, or significant, recovery of data might be practicable, but it is worth noting that discussing the problem with a RAID data recovery expert prior to embarking upon a RAID rebuild is a good idea if the data has importance.
Failed Hard Drives
Where there are multiple hard drive failures reported a data recovery operation is the only approach that might give access back to the data. RAID5 will operate in degraded mode if one disk has failed, if two more more fail then the RAID will fail. One important factor is whether the all failed drives stopped working at the same time or whether one had failed some time previously and the unit had been running for a while with a missing disk. Thereafter the data recovery process must concentrate on the hardware and whether any or all of the failed drives have actually stopped working, developed a number of errors, or are actually fine and there is a controller problem.
It is surprising how often during a RAID data recovery diagnosis that we find problems with only one drive though three are reported as failed, quite often we find there are not problems with any of the drives. It appears that some systems respond to electronic problems or instabilities on the "bus" to which the drives are attached by taking drives off-line and asserting error conditions.
If you have a problem with a RAID5 it is essential that you seek professional advice. With RAID hardware you do not get much control over what is happening and any steps taken in an attempt to rectify a problem could be a one-way trip to data oblivion. There is no way back from an erroneous rebuild.
Last Updated (Wednesday, 22 July 2009 09:17)