What is RAID?
We will be adding some more content on this page to give a clear definition of RAID, and to help you understand RAID Levels / Types and to choose which configuration best suits your IT system.
For now, you can read an introduction on our RAID Recovery page and find out about Levels below.
If you are still hungry for learning about RAID history and theory you could read this article at Webopedia.
There are multiple types of RAID Systems, each one offers a specific level of redundancy, performance or capacity. These different types of arrays are known as Levels. Data is distributed (striped) across the drives dependent upon the desired balance between performance, fault tolerance and backup required.
Redundancy is provided by copying data across two or more drives – known as mirroring, which helps to preserve data if one drive fails. For performance benefits a technique called “striping” provides reading or writing to more than one disk at the same time. This will help to increase overall performance. This method allows data to be broken into sections which are sent to various drives in the array and thus speeding throughput.
Level 0 is usually referred to simply as RAID. Assuming there are three drives, instead of writing everything on the first disk, the data will instead be split up; eg one portion on drive one, the next on drive two and the final on drive three.
The files have been broken down into pieces and instead of reading from one drive, the data is now read in parallel from all drives and then combined by the controller. By using all three drives, the access to the file is vastly improved. Performance is improved because files are being read from at least 2 drives at the same time. In theory Level 0 can work with one drive but this rarely happens in practice.
If one of the disks fails, the entire file is ruined due to the lack of coherent, consistent data. Level 0 does not provide any redundancy. A sudden RAID failure can result in being unable to access critically important information. Although this provides for excellent performance, the lack of redundancy creates issues for any critical system.
Few RAID devices on the market today offer only Level 0, however due to the expense of SSD drives it has become popular to configure 2 of these as Level 0 to increase capacity.
Level 1 or mirroring is the process of using two drives and placing identical data on both. If one drive fails, the data is safely retrievable from the second. Level 1 is the most basic form of true RAID. Although most RAID 1 arrays have two drives, some systems have additional drives for additional fault protection. Businesses often mirror RAID 0 or RAID 5 with a separate Level 1 array, creating an identical copy of the primary array to add additional redundancy.
Level 1 is an extremely dependable option, but does not offer performance improvements over a standard hard drive. RAID 1 is typically used when high fault tolerance is more important than performance.
Level 0+1 provides performance benefits with built in redundancy. It offers ideal functionality for storage servers, web servers and other types of high-performance systems. The segments of the system are individual RAID 0 arrays, mirrored to provide redundancy. If a drive fails, the system will operate effectively until the damaged drive is replaced. This is one of the most popular configurations and most configurations at higher levels offer at least this level of Performance and redundancy, by adding additional drives or additional arrays.
RAID-3 AND RAID-4
Level 3 data is striped across multiple disks at a byte level. The parity information is sent to a dedicated parity disk, but the failure of any disk in the array can be tolerated. The dedicated parity disk generally serves as a performance bottleneck, especially for random or short writes, because it must be accessed any time anything is sent to the array.
Level 4 similar to RAID 3, but parity is checked at the block level. The difference is very minor, and is only applicable when you start looking at the finer art of RAID systems. For each row of written data, any recorded block is written on a parity disk. Level 4 does not use distributed parity. RAID 4 requires at least three disks for complete implementation and configuration. Level 4 works well with large files – when sequential read/write data is used.
Level 3 and 4 are very rarely used and are best suited where large files are continuously read or written for a long period of time eg video files.
Level 5 is also similar to RAID 3 and is the most popular option. In RAID 5, the last disk is not the only drive that contains parity data in the array. In RAID 5, you rotate the parity in the five disk drives. If one disk dies, you can recreate the data using the remaining four disks. This option provides good performance and good redundancy and helps to increase data availability. Level 5 is an ideal option for web servers and storage servers.
RAID 5 allows for faster data access speeds than RAID 4 because RAID 5 can write to all of the disks at the same time. RAID 5 systems need to be rebuilt after losing a hard disk. It is at the rebuild stage when it is common for a second disk to fail when the RAID controller stops functioning properly.
Level 6 is similar to RAID 5 in terms of striping, but is different because RAID 6 can tolerate two discs failing. Because RAID 6 systems need to perform twice the parity calculations, write speeds are often limited compared to RAID 5.
Level 10 is a nested RAID configuration that provides the redundancy of RAID 1. It is known as “stripe of mirrors” and can handle one hard drive failure per set without losing data. When drive failure occurs, the damaged member should be replaced as soon as possible. They are highly redundant and extremely reliable and is best suited to mission critical applications.
Level 50 consists of two or more RAID 5 Arrays with data striped across each array to form a RAID 0 and offers better fault tolerance and faster read/write speeds without the single point of failure of a solitary RAID 0. These are typically used for mission-critical applications. Level 50 arrays can typically sustain multiple hard drive failures, however if 2 disks fail in the same array you will have total data loss.
Level 60 is to Level 6 what Level 50 is to Level 5 – ie if offers the same benefits as RAID 50 but has the additional resilience that two disks in the same array can fail without data loss.
When is RAID 60 Overkill? – TechRepublic.com
For an in depth explanation of RAID see IBM Understanding RAID Technology.