Non-redundant systems

The term RAID is incorrect, however is still applied to RAID 0 (RAID level 0, Stripe) or JBOD (Just a Bunch Of Disks, Span) disk arrays. The practical use of these systems are:

• JBOD: extend storage with more disks, even those with different size. Each component of JBOD follows previos to provide monolithic storage with size, equal to sum of component sizes. JBOD is supported by most hardware RAID chips as well as with software (e.g. Dynamic Disks under Windows can span among different disks or disk partitions).
• RAID0: stripe set on disks of equal size. The data on RAID0 is divided to 'stripes' of equal size and cyclically distributed among all disks. Such 'stripe' size is usually from 512 bytes and up to 256KB. The purpose of data striping is to distribute long data fragment among all disks. This allows to issue data exchange requests to all drives the same time an to speed-up much these operation with parallel read or write. These systems are fastest and utilize all disk space.

Perspectives of data recovery from these systems are obvious: even if one disk drive from RAID could not be read - the data from this component could not be recovered. If such kind of failure occurred on JBOD - entire fragment of span could not be recovered anymore. For RAID0 this will affect all data on RAID (e.g. if RAID0 is built on 4 disks with stripe size 16KB, after single disk failure RAID will have 16KB 'hole' after each 48KB block. In general, this would mean any file with size over 48KB cannot be recovered).
If one or more disks from RAID0 or JBOD failed - stop use the disk immediately and bring it to data recovery laboratory. Only physical drive repair could help to recover the data!

In case system has problems, not related to disks (e.g. controller settings were re-set, controller failure or damage etc.) - it's possible to recover data even in case of file system logical damage. It's only required virtually to assemble the original monolithic storage with data recovery software. To do this you have to specify what disks are included into RAID, what the drive order was and what stripe size for RAID0 was. The data recovery software will read data from components in the same manner like it does RAID controller so it will be able to access good files on virtually reconstructed RAID.


Mirrors

The classic 'mirror' implementation is called RAID1. Each component of such RAID contains the same data so information could be recovered from any good component of the RAID. For this kind of RAID controller may still do parallel reading operations to speed-up files read access. This kind of storage is most redundant and has the best perspectives of data recovery.

Usually you do not need to do other actions with data recovery software apart from just direct, single component read.


Advanced redundant systems

These systems are the compromise between high speed of storage access, storage size and redundancy. Usually they use idea of striping from RAID0 but on-disk data is extended with extra information (the parity information) that adds redundancy and allows recovering data or even continuing RAID operations after component failure.

Among these systems are RAID3, RAID4 or RAID7 (stripe set with dedicated parity), RAID5 (stripe set with distributed parity) and RAID6 (stripe set with double distributed parity). The 'single' parity means RAID data could be recovered or system can operate after single component failure; 'double' parity - after up to two components failure.

RAID3 and similar systems use classic RAID0, extended with one more disk to store parity. RAID5 and RAID6 distribute parity among all disks to speed-up parity update process for data write operations.

Data recovery from these systems is possible for complete RAID, as well as if one (for RAID3, RAID4, RAID5, RAID7) or up to two (for RAID6) components could not be read. If more disks failed - stop use the disk(s) immediately and bring it to data recovery laboratory. Only physical drive repair could help to recover the data!

If data recovery is possible without drive repair, just assemble the RAID in data recovery software by specifying drives (including placeholders for a missing drive), drives order, stripe size and parity distribution algorithm. The data recovery software will read data from components in the same manner like it is done by RAID controller so it will be able to access good files on virtually reconstructed RAID.


Hybrid systems

Vendors often use RAID-on-RAID configurations to improve overall performance, add redundancy and so on. Usually such systems are combinations of RAID systems mentioned above. Most classic are systems like RAID10: this is several 'mirrors' with 'stripe' over them. Mirrors here ensure redundancy and stripe over mirrors adds read/write speed. The data recovery from this system is quite simple: take any good component from each mirror and virtually build RAID0 over it.

The more advanced systems are RAID50 (stripe over RAID5), RAID51 (mirror of RAID5) and so on. To reconstruct such system, just build each component of RAID (in this example - each RAID5) and for RAID50 - build RAID0 over set of RAID5.

Our software supports virtual reconstruction of complex RAID systems of any multiplicity.