|
Understanding RAID,
its three main permutations, and of the three, which type (also called RAID "level") is used in your particular application, is of importance that simply cannot be overstated. As providers of data recovery service, we at MicroCom are often witness to aggravation of emergency situations stemming from incomplete IT system operator understanding about how hard disk drives within their data storage hardware are actually being used. At such times of data center crisis, misconceptions of actual RAID implementation leads frequently to remedial actions by the operator/user that make matters even worse.
RAID-0, pronounced "raid zero", is aptly named & defined as a non-redundant group of striped disk drives without parity. The stripes are small segments of file data, dispersed in sequence to each drive in the array. In other words, different parts of every file are recorded on different drives. If one drive in a RAID-0 array crashes, the entire array crashes. RAID-0 provides the highest performance of all RAID levels: it is the fastest and yields the greatest storage capacity efficiency of all hard disk RAID arrays. However . . .
With this array configuration there is considerable danger with respect to the accessibility or "availability" of your data; it provides zero fault tolerance!
:: For more in-depth explanation & discussion, see RAID DEFINITIONS ::
RAID-1, also known as "mirroring" is simply a pair of disk drives which store duplicate data, but appear to the computer as a single drive; the number of drives in the array will always be an even number. Some RAID-1 configurations may improve system read or data access performance and leave the write performance unchanged, but others can leave read performance unchanged while degrading or slowing the write performance. It is the only choice for fault tolerance if no more than 2 drives are used.
While RAID-1 makes the least efficient use of drive storage capacity, for environments where data storage simplicity is important and data availability is critical, it is the RAID array of choice.
RAID-2, RAID-3, and RAID-4, are RAID configurations that were defined decades ago when the earliest concepts for increasing data availability by means of data redundancy were first being developed. In those days computer performance, and especially disk drive performance was much slower than it is today. In that era these configurations made sense for optimizing system response for certain applications. Today they are rarely if ever used.
RAID-2 arrays sector-stripe data across groups of drives, it offers no significant advantages over RAID-3. While both RAID-3 and RAID-2 sector-stripe data across groups of drives, with RAID-3 one drive in the group is dedicated to storing parity information and relies on the embedded ECC in each sector for error detection. RAID-3 arrays cannot overlap I/O operations, thus it can give good performance only in single-user, single-tasking environments with comparatively large files.
RAID-4 allows read operations to be overlapped, but since all write operations must update the parity drive and cannot be overlapped, elapsed times for storing or editing data is penalized.
RAID-5 or "rotating parity array" avoids the write bottleneck caused by the single dedicated parity drive of RAID-4. In contrast, each drive takes turns storing parity information for a different series of stripes in a rotating fashion so that the parity is distributed among all member drives in the array. RAID-5 is today the most commonly implemented type of fault-tolerant data storage disk drive array, offering optimal performance combined with high storage capacity efficiency. The larger the number of array elements (i.e., hard drives), the more that storage efficiency is increased; the array may theoretically contain any number of drives, but more than nine or ten elements is extremely uncommon, usually there are less than six. It takes a minimum of 3 drives to create a RAID-5.
RAID-5 has always been the best choice in multi-user, data-processing environments which are characterized by high read access of data and a comparatively low need for write operations.
With today's hardware operating literally thousands of times faster than technology permitted when RAID designs were in their early days of use, any performance penalties are now miniscule and easily outweighed by the assurances and fault-tolerance provided by data redundancy. For very good reasons then, among the server farms and clustered domains in the world of information technology, RAID-5 has become by far the most popular configuration of all data storage subsystems.
SUPER RAIDs
They are called "super RAID's" because an increased measure of performance and/or fault tolerance, also known as data availability, is designed into these configurations by means of striping and boosted redundancy. In turn, this is alternatively achieved through use of added parity, two combined or nested elemental RAID arrays, and mirrored sets.
RAID-6 can tolerate the failure of up to two simultaneous array member hard drives without downtime or data loss. This capability is achieved by recording two independent parity algoritms and uniformly distributing each to all hard drive members of the array. RAID-6 requires a minimum of four drives. It is similar to RAID-5, but has one additional drive with a RAID controller that creates an added layer of redundancy. In other words, it has doubled parity in each stripe to produce a doubly fault-tolerant data storage subsystem.
RAID-10, the combination of the numerals 1 and zero, where two identical RAID-0 arrays are mirrored to produce a fault-tolerant, high-performance data storage subsystem. This configuration provides the simplest example for what is meant by "nesting", where the RAID-0 array is nested within a RAID-1 array.
RAID-5x, these are various RAID levels constructed by nesting RAID-5 arrays within deeper, more complex structures. RAID-50, RAID-51, and RAID-53 are examples. Researchers delving into this area will encounter some discrepancies in regard to exactly what the proper nomenclature for some of these configurations are or should be; some are trademarked, proprietary expressions. Here again, a good conceptual understanding of all can be derived from sound knowledge of what we call the three fundamental RAID levels: zero, one, and five.
|
