RAID

Servers and network storage rarely store data on a single hard drive. Usually several drives work together as an array. This increases fault tolerance or speed – and distributes the data across many drives. When dismantling, every single one of them must therefore be taken into account.

How RAID works

The concept was described at the University of California in Berkeley in the late 1980s. A RAID controller or software in the operating system distributes the data to the participating drives according to fixed rules. To the operating system, the array appears like a single drive. If one drive fails in redundant levels, the data continues to run and is restored after replacement.

The most important RAID levels

Level	Principle	Minimum number of drives	Tolerates failure of
RAID 0	Striping: data is distributed block by block	2	No drive
RAID 1	Mirroring: data is mirrored	2	One drive per mirror
RAID 5	Striping with distributed parity	3	One drive
RAID 6	Striping with double parity	4	Two drives
RAID 10	Combination of mirroring and striping	4	One or more, depending on distribution

Important: RAID is not a substitute for data backup. Deleted or encrypted files, for example after an attack, are affected on all drives of the array.

What to note when retiring devices

• Every drive counts: Even with RAID 0 or RAID 5, every drive contains contiguous data blocks from which files or parts of them can be reconstructed. With RAID 1, a complete copy resides on each drive.
• Erasure across the array or per drive: If only the logical drive is erased, areas outside it, such as reserve and metadata areas, may remain. Erasing each individual drive is safer.
• Spare drives and replaced drives: Hot spare drives and drives replaced earlier belong on the inventory list as well.
• Controller cache: Some RAID controllers have their own memory. Clarify whether data can remain there.

In storage networks, data is often managed as logical units. This is matched by LUN erasure. How the dismantling of entire server rooms works is described in the guide Selling Used Servers and Dismantling a Data Center.

RAID systems at Second IT

Second IT erases drives from servers and storage systems with Blancco, including in storage environments at the level of LUNs. Every drive is recorded by serial number and receives an erasure certificate. Defective drives are destroyed in-house in accordance with DIN 66399 (up to H-5 and E-5). More on certified data deletion.

Frequently asked questions about RAID

Which RAID level is the best?

That depends on the goal. RAID 1 and RAID 10 offer high safety and good performance at the cost of more capacity. RAID 5 and RAID 6 use capacity better but reconstruct more slowly after a failure.

Is data on a single drive from a RAID 5 readable?

Usually not whole files, but data blocks yes. Depending on block size and file type, content can be partially reconstructed. That is why every drive must be erased or destroyed.

Do I have to dissolve a RAID before handing it over?

No. For erasure, the drives are usually addressed individually. What matters is a complete list of all drives with serial numbers so that each one receives a record.

Does a RAID replace a data backup?

No. A RAID protects against the failure of individual drives, not against deleting or encrypting files. If files are deleted or, for example, encrypted after an attack, all drives of the array are affected. A data backup therefore remains necessary.

What is the difference between RAID 0, RAID 1 and RAID 5?

RAID 0 distributes data block by block across at least 2 drives and tolerates no failure. RAID 1 mirrors data across at least 2 drives. RAID 5 uses distributed parity across at least 3 drives and tolerates the failure of one drive.