It is created as mirror sets with striping going across the mirror sets. Whether or not a second loss will cause a loss of data depends entirely upon where the second disk loss occurs. If you lose one disk in each mirror set, you will have no problem. However, if you lose both disks in one mirror set, the whole volume will be lost.
Each pair of disks becomes a mirror set. Data is written in strips across each of the pairs. You sacrifice fifty per cent of available space to receive fast write performance along with failure behaviour that can be understood simply.
Data is striped across mirror sets after the disks in the system are paired into mirror sets. For example, four disks create two mirror sets. The data being written to the disks is spread out across the two sets of mirror disks for speed. Each member of a mirror set contains an identical copy of the other member's data for safety.
It really does matter what order they are applied in. This order is often misinterpreted. RAID 10 is not striped-then-mirrored (RAID 01), which has quite different behaviour under failure. In RAID 10, if a single disk fails, you lose no data on your RAID set because its mirror holds the exact same data.
The outcome of this is that RAID 10 can survive two or more disks failing at the same time if those disks fail in different pairs. Lose one disk per pair, and you still have your array up and running. If both halves of the same pair fail, then you've lost the whole volume, regardless of how many other drives are still functioning.
Both halves of one pair failed. You will only lose an entire RAID 10 when both halves of one pair of mirrors fail. This is due to the fact that most mirroring systems use identical disks purchased together for their mirror sets. These disks are also installed together and wear out similarly.
The first failure went unnoticed. When you experience a single failure (the first), you don't notice it, as a degraded array appears to function normally until the second failure in the pair occurs months later. At that point the RAID 10 volume drops without warning.
A rebuild ran onto the wrong member. If a rebuilding operation were to occur on the wrong side of the pair, that would overwrite the good side. However, recovering from such a situation is possible, but far more difficult than it should be.
The controller lost its configuration. The disks themselves are perfectly fine, but the configuration map that describes which disk belongs to which RAID pair is gone. It's possible to recover the configuration information, but it won't be possible to read the information contained on the disks until that configuration has been recovered.
A minimum of one failure or potentially up to fifty per cent of the total number of members, but only if those members fail in different pairs. The failure of two members within the same pair will destroy the volume regardless of how many other disks have survived.
Different, not safer. RAID 10 rebuilds much quicker as it copies over one entire disk rather than having to recalculate the parity. This results in a significantly shorter “risky” window. RAID 6 can withstand the loss of any two members, regardless of where the members are located.
When this happens, then both of the failed members were located in the same pair. Send all members with labels based upon their respective bay locations: the surviving pairs are intact, and often at least part of the failed pair can be recovered. In some cases, this may be sufficient.
Yes. Membership of a pair and the geometric pattern of stripes are determined from the physical disks contained within your array. This is a standard procedure for any array whose controller has gone.
From £500 +VAT, fixed in writing after completion of our complimentary diagnostic service, regardless of the number of members in your array.
A restoration process takes anywhere from five to ten working days. Every single member must first be completely imaged prior to reconstruction, and this imaging is most of the actual time spent during a full restoration.