EN 
11.09.2024 WELCOME IN MY WORLD

This website is originally written in the Czech language. Only part of the content is machine (AI) translated into English. The translation may not be exact and may contain errors.

Tento článek si můžete zobrazit v originální české verzi. You can view this article in the original Czech version.
RAID - Redundant Array of Independent Disks

RAID - Redundant Array of Independent Disks

| Petr Bouška - Samuraj |
In the field of data storage (computer storage), RAID is a key term. This is a technology for storing data over several disks. There are a number of different methods (schemes), but in most cases reliability and speed increase. In general, we can say that RAID combines several disks into one logical unit. Data is usually split and stored (and read) on several disks at once, thus increasing write/read speed. Furthermore, a checksum is stored on the next disk, so that data can be restored even if a certain part (disk) fails.
displayed: 12 227x (12 214 CZ, 13 EN) | Comments [2]

RAID can be created software-wise using operating system resources, which is cheap but less reliable and efficient. Or hardware-wise, where everything is managed by the controller. Nowadays, even most home computers are equipped with a controller with simple HW RAID, so it's not a problem to use RAID 0 or RAID 1.

Basic types of RAID

RAID 0 - striping
Minimum 2 disks, data are stored alternately in so-called strips. Simultaneous writing/reading to all disks (each time something different), both are accelerated n times. Failure of one disk means losing all data.
RAID 1 - mirroring
Minimum 2 disks. Reading from all at once, i.e., acceleration, the same data is written to both disks. Safe, but only part of the capacity is used.
RAID 3 - striping with parity
Data is added with a checksum on a separate disk. Replaced by RAID 5.
RAID 5 - striping with distributed parity
Minimum 3 disks. Data + parity are stored, from which damaged data can be recovered. Reading from all disks at once except parity, i.e., acceleration n-1 times. When writing, parity needs to be calculated - HW acceleration on the controller, used capacity is (n-1) times the disk size. Can handle the failure of one disk.
RAID 6 - striping with double distributed parity
Similar to RAID 5, but with double parity, can handle the failure of two disks at once, minimum 4 disks, capacity (n-2) times.
RAID 10 - RAID 1 and RAID 0
Speed of RAID 0 and safety of RAID 1, minimum 4 disks, capacity is half, faster than RAID 5.
RAID 50 - RAID 5 and RAID 0
Minimum 6 disks, similar to RAID 5, faster.

Related terms

hot-swap disk
feature of a disk array and disk that allows disk replacement while running
hot-spare disk
spare disk that waits for a disk failure and automatically replaces it, in some systems the hot-spare function is automatically cycled (due to load)
Author:

Related articles:

Computer Storage

Data storage is a vast and complex issue in the computer world. Here you will find articles dedicated to Storage Area Networks (SAN), iSCSI technologies, Fiber Channel, disk arrays (Storage System, Disk Srray) and data storage and storage in general.

If you want write something about this article use comments.

Comments
  1. [1] kulda

    Muzu se zeptat jaky RAID se nejcasteji pouziva ve firemni siti?

    Thursday, 06.09.2018 08:46 | answer
  2. [2] Samuraj

    respond to [1]kulda: Záleží na účelu. V serveru se třeba doporučuje jiný RAID na systémových discích a jiný na datových. Dnes máme většinou vše na poli, kde děláme velký RAID přes hodně disků. Praxe asi každému ukáže, že RAID s možným výpadkem jednoho disku je nedostatečný (často při rebuildu odejde další disk, protože je hodně zatížený). Takže dost se používá RAID 6. Nebo někteří výrobci, jako NetApp, mají vlastní RAID, který doporučují.

    Thursday, 06.09.2018 09:04 | answer
Add comment

Insert tag: strong em link

Insert Smiley: :-) ;-) :-( :-O

Help:
  • maximum length of comment is 2000 characters
  • HTML tags are not allowed (they will be removed), you can use only the special tags listed above the input field
  • new line (ENTER) ends paragraph and start new one
  • when you respond to a comment, put the original comment number in squar brackets at the beginning of the paragraph (line)