Posted on

3 biggest mistakes people make in RAID recovery

1. Doing RAID recovery where there is no need to recover RAID

Some people do not know at all whether they use RAID or not. This often happens when a user heard somewhere about RAID technique and arrays, and now he or she thinks that there is a RAID on his or her system. And then, if something fails, these users start recovering their “RAID” even consisting of only one drive. Some may find it hard to believe, but in our statistics it is one of the most common mistakes users made in RAID recovery.

People also often can’t tell a RAID failure from a filesystem failure and apply RAID recovery to the cases where filesystem recovery is needed. In such situations, RAID is working properly while data corruption is caused by something else, for example a filesystem driver problem. Data recovery in these cases is no different from restoring data off a single drive and typically is solved by any data recovery software.

Start with file system recovery if:

  • you have deleted a file on your RAID system and then cannot find it in the Recycle Bin;
  • Windows reports RAW file system for the volume managed by RAID;
  • Windows doesn’t start. It makes sense to add one more disk to the system, put a temporary Windows installation on it not touching the RAID, and then proceed to investigate the condition of the array data using this temporary installation.

Start to recover RAID if:

  • RAID-controller displays the error message and refuses to bring the array online;
  • NAS device doesn’t turn on;
  • RAID array in question is no longer displayed in the list of arrays in a RAID-controller software or in a NAS control panel;
  • failure has occurred immediately after some action with RAID, for example, after disk replacement in a RAID5.

2. Wrong RAID type

Very often even the most advanced users confuse different RAID types. RAID0 and JBOD are most often confused; however, still relatively large number of errors is the share of RAID5 with RAID10. Although RAID recovery software can tell RAID10 from RAID0, it is unable to distinguish a RAID5 with one disk missing from a RAID10.

However, it should be noted that all RAID recovery software is solely read-only which means that if you confuse the array type, nothing bad will happen. Surely, with the wrong RAID type you do not get the correct RAID configuration. But once you specify the correct RAID type, you will have a good chance of getting a solution for the array.

To avoid this mistake first you should exclude impossible types of RAID following the rules like:

  • RAID cannot be created on one disk; therefore, if all you have is a single disk, it is for sure a non-RAID system.
  • JBOD (Just a Bunch Of Disks, also called Span) cannot be ruled out if you have multiple disks.
  • If all you have is two disks, it can only be RAID0, RAID1, or JBOD.
  • Three disks cannot be used for a RAID 6 or RAID 10.
  • RAID controllers built into common motherboards, like ICH-R series controllers, typically do not support fancy or computationally heavy RAID levels, like RAID 1E or RAID 6. Check the motherboard manual to see what levels are supported.

*Note: Missing disks are not taken into account in all the above considerations.

If you know for sure what the array capacity was, you can estimate what RAID types match to what capacity:

  • capacity of a RAID 0 and JBOD is equal to the sum of member disks capacity;
  • RAID 1 is usually created on two disks and total RAID 1 capacity is equal to the capacity of one disk;
  • in case of a RAID 5 of N disks, the capacity is (N-1) capacities of separate disks; for a RAID 6 is (N-2), where N – the number of disks.

Then, you should go over all the remaining possible types using different modes of data recovery software.

3. Rebuilding a RAID with wrong parameters

If you deal with a fault tolerant array that can survive one or multiple disk failure, then you should be familiar with a rebuild. It is a standard procedure that is typically performed after replacing a failed disk in RAID5, RAID10, or RAID6.

Some people try to force the array online after a problem has occurred, using either default parameters or their best guess about the previous setup. However, there is a catch – some controllers will start a rebuild if the array is forced online without asking you first. This means you have only one chance of doing things right.

If an array was rebuilt using wrong parameters, then in most cases it is impossible to recover data off it. The speed at which an array rebuilds is about 50-100 MB per second. If the data on your array was managed by NTFS which is known to store file metadata (MFT) along the first 2-4 GB of RAID, then after about 30-90 seconds the filesystem records will be significantly damaged. Thus, even if you cancel the synchronization within minutes after the start, you still lose half of the files. Neither data recovery software nor a data recovery lab can help you in such a situation. In case of the fully completed rebuild, the array contains nothing even remotely resembling the previous data.

The only thing that can help to avoid this mistake is to back up all the array member disks before rebuilding the RAID. Alternatively, you can just copy the important data using read-only software.

 

Written by Elena Pakhomova of www.ReclaiMe.com, specializing in data recovery solutions for various NAS (Network Attached Storage) devices.

Posted on

STEM degrees

“Every year U.S. schools grant more STEM degrees than there are available jobs. When you factor in H-1B visa holders, existing STEM degree holders, and the like, it’s hard to make a case that there’s a STEM labor shortage.” Even in the IT industry, which employs the most tech workers and is expected to experience the most growth over the next decade, not everyone who wants a job can find one. Anecdotal evidence, the article points out, is piled high on the side of there being a glut instead of a shortage. “If there was really a STEM labor market crisis, you’d be seeing very different behaviors from companies,” Ron Hira, an associate professor of public policy at the Rochester Institute of Technology, in New York state told IEEE Spectrum. “You wouldn’t see companies cutting their retirement contributions, or hiring new workers and giving them worse benefits packages. Instead you would see signing bonuses, you’d see wage increases. You would see these companies really training their incumbent workers.” In a related opinion piece, “Is a Career in STEM Really for Me?” an 8th grader ponders her options, and finds science and engineering far down on the list.