Close
    logoMevspace Docs
    English

    What is a RAID and which should you choose?

    Reading time: 12 min

    "RAID" stands for Redundant Array of Independent Disks. Even though this name may sound a bit confusing, the RAID is quite simple to understand and use. As you probably noticed, mostly whenever a RAID is mentioned, it has a number - RAID 0, RAID 1 - up to RAID 6. In this article, we'll explain what the RAID is and how its level differ from each other.

    To set up a RAID you need to combine two or more disks; this allows you to achieve data redundancy, better efficiency (faster data writing/reading), or both. Such setup provides additional possibilities that cannot be achieved through traditional disk setup. RAID has multiple levels, distinguishable by numbers.

    It is also important to know that a RAID array is not a backup system (although it can be used to increase data safety). It can, however, have internal security measures like data mirroring or various IT and/or mathematical solutions that allow data recovery. It is also worth mentioning that it operates interdependently of disks outside it.

    There are two types of RAID, the difference being the method of installation:

    • Hardware RAID uses a hardware controller that takes over the array's disk management and thus relieves other computer components, most notably the processor.

    • Software RAID does not require any physical components, and instead, is set up in the computer's BIOS/UEFI or in the operating system. However, unlike hardware RAID, it uses up processor resources.

    Mevspace's (formerly Skynode) dedicated servers. can install an operating system automatically, should the user so desire. During this process, you can select between either RAID 0, RAID 1, or none. Below, you will find their descriptions which will help you learn about them and know whether you should use a RAID array or not.

    What is RAID 0?

    RAID 0 makes the drives work as one - one file is divided between all of the array's disk drives. In other words, all of them participate in writing and/or reading of a single file, each one operating on a portion of it.

    skynode raid 0

    Fig.1 - Storing "Skynode" data in RAID 0

    RAID 0's biggest upside is increased efficiency and, unlike in the case of RAID 1, increased disk space. It's important to note that it is only as fast as the slowest drive - if fast and slow disks are installed together, not only they won't work much faster, but the slower disk will slow down the faster one.

    Its biggest con is data unrecoverability - if a failure happens, all of the data stored in the array is going to be lost. Even if only one disk fails, the data lost is a part of files on other drives, and thus, they also become useless. That's why it is highly recommended to often backup RAID 0 arrays.

    It's best used for storing data like the operating systems or game files. In their case, better write/read speed will greatly improve the machine's efficiency and user's experience; and most of the time, they can be easily re-acquired (e.g., through a removable drive or download). At the same time, since it's impossible to recover data from a RAID 0's failure, it's highly discouraged to keep personal data like documents, projects, photos, or movies. Such files, which don't require such high operating speed, are best stored on independent HDD drives - a much slower, but much safer alternative.

    For users who chose a RAID 0, we strongly recommend doing regular backups, as recovering data from failure is impossible.

    What is RAID 1?

    In RAID 1, disks are "mirror images" of each other - such setting is called mirroring. Unlike in RAID 0, disks do not split data between them, but copy it instead.

    skynode raid 1

    Fig.2 - Storing "Skynode" data in RAID 1

    RAID 1 is used because of the safety it offers: even if a disk fails, all of the data stored in this array will remain fully accessible. For example, if in a RAID 1 array composed of 4 disks 3 of them fails, all of the data will remain intact - all of it is still stored on the 4th, working drive.

    Its biggest con is the constant disk space - no matter how many drives are installed, it will remain as same as the smallest one. Since drives do not cooperate in data writing/reading, RAID 1 is slower than RAID 0.

    While it is technically possible, there's no need to install more than two drives in a RAID 1 array - chances for both of them failing at the same time are close to zero.

    It is used to ensure the high safety and recoverability of data stored. Important projects, personal files, and such are often stored in a RAID 1 array.

    Other RAID levels

    There are many other RAID types. Among them are, for example, arrays with disks used for storing special codes and keys that are used to recover data in case of failure. Unlike RAID 1, one of those "anti-failure disks" holds data corresponding to many "storage" disks.

    The higher RAID levels are as follows:

    • RAID 2: Data is striped between some of the disks. Unlike RAID 0 and RAID 1, RAID 2 has additional security disks, containing error-correction data. Because of that, the RAID 2 is very slow, but in case of a disk failure, it can be rebuilt by the rest. Nevertheless, this RAID level is nowadays obsolete and unused.

    • RAID 3: The data is split like in RAID 0. However, there's another disk drive containing control sums. This solution is immune to a single disk's failure, but it's slow and time-consuming, thus not widely used.

    • RAID 4: This array is like the one described above and likewise, it is immune to a single disk failure thanks to a special parity disk. The difference is that the data is split into blocks of a certain size, a power of 2 (e.g., 16 kb, 32 kb, 128 kb, etc.). This solution makes the array operate fast on large files, but very slow on smaller ones. It remains, however, more efficient than RAID 2 and RAID 3.

    • RAID 5: Like in the case of RAID 4, parity bits used for recovery are present. However instead of being kept on a dedicated drive, they are spread on all disks in the array. Reading operations are as fast as in RAID 0. However, without a hardware controller, writing is significantly slower.

    • RAID 6: It's very similar to RAID 5, except one "set" of parity bits is stored twice - on two disks simultaneously. This RAID has higher costs than the others, but it's immune to the failure of two disks instead of just one.

    Solutions from different RAID levels can be used together. The structure of arrays created in such ways reminds a tree.

    • RAID 0+1: RAID 1 is the base of this array. However, instead of two disks copying the same data, it is two RAID 0 arrays copying each other.

    • RAID 1+0 (RAID 10): Its basic premise is the same as the RAID 0. However, disks in this array have "twins", working just like the RAID 1.

    Are RAID arrays worth using?

    Using the RAID is not necessary. Most home and office users' computers have single disks, not set up into any array. While all of the RAIDs have their pros, they also have their cons. With RAID 0, the drives are dependent on each other, meaning the data is unrecoverable if a failure occurs. In the case of RAID 1, it is limited disk space. If you don't need faster write/read and you don't need a constant backup, using RAID is not recommended, and can cause you additional trouble.

    An interesting fact is that SSDs can be said to be RAID 0 themselves. As they are made out of an array of cooperating memory chips.

    Which RAID array should I choose?

    Whenever deciding on a RAID - or lack of it - one should consider its pros, cons, and uses described above. If the machine is needed to work as fast as possible, RAID 0 is the best solution - given that all documents will be stored on the external drive. However, if it is not speed but safety that's most important - RAID 1 will be the best choice. If you don't need a RAID, you can completely ignore it and install your disks in a standard unconnected state.

    RAID arrays in Mevspace

    At this moment, our servers allow setting up either RAID 0, RAID 1, or none. Applying the chosen array is simple and can be done by using the reinstallation option in Mevspace's Control Panel or when purchasing a dedicated server. In the future, Mevspace(formerly Skynode) will also offer dedicated servers that will allow the user - given they have the right amount of drives - to set up any other RAID, both software and hardware, with proper equipment.

    Previous
    TCP protocol
    Next
    What is the BGP protocol and how it works?