| |
|
The "RAID" acronym first appeared in 1988
in the earliest of the Berkeley Papers written by Patterson,
Gibson & Katz of the University of California at
Berkeley. The RAID Advisory Board has since substituted
"Independent" for "Inexpensive". A series of papers written
by the original three authors and others defined and
categorized several data protection and mapping models for
disk arrays. Some of the models described in these papers,
such as mirroring, were known at the time, others were new.
The word levels used by the authors to differentiate the
models from each other may suggest that a higher numbered
RAID model is uniformly superior to a lower numbered one.
This is not the case.
RAID 0 (Striping)
RAID 0: Striped Disk Array without
Fault Tolerance
RAID Level 0 requires a minimum of 2 drives to
implement.
RAID Level 0 is a performance oriented striped
data mapping technique. Uniformly sized blocks of storage are
assigned in regular sequence to all of an array's disks. RAID
Level 0 provides high I/O performance at low inherent cost. (No
additional disks are required). The reliability of RAID Level 0,
however is less than that of its member disks due to its lack of
redundancy. Despite the name, RAID Level 0 is not actually RAID,
unless it is combined with other technologies to provide data and
functional redundancy, regeneration and rebuilding.
Advantages: RAID 0 implements a striped disk
array, the data is broken down into blocks and each block is
written to a separate disk drive. I/O performance is greatly
improved by spreading the I/O load across many channels and
drives. Best performance is achieved when data is striped across
multiple controllers with only one drive per controller. No parity
calculation overhead is involved. Very simple design. Easy to
implement.
Disadvantages: Not a "True" RAID because it is
NOT fault-tolerant. The failure of just one drive will result in
all data in an array being lost. Should never be used in mission
critical environments.
Recommended Applications: Video Production
and Editing; Image Editing; Pre-Press Applications; Any
application requiring high bandwidth.
[ TOP ]
RAID 1 (Mirroring)
RAID 1: Mirroring and Duplexing
For
Highest performance, the controller must be able to perform two
concurrent separate Reads per mirrored pair or two duplicate
Writes per mirrored pair.
RAID Level 1 requires a minimum of 2
drives to implement.
RAID Level 1, also called mirroring, has been
used longer than any other form of RAID. It remains popular
because of its simplicity and high level of reliability and
availability. Mirrored arrays consist of two or more disks. Each
disk in a mirrored array holds an identical image of user data. A
RAID Level 1 array may use parallel access for high transfer rate
when reading. More commonly, RAID Level 1 array members operate
independently and improve performance for read-intensive
applications, but at relatively high inherent cost. This is a good
entry-level redundant system, since only two drives are
required.
Advantages: One Write or two Reads possible per
mirrored pair. Twice the Read transaction rate of single disks.
Same write transaction rate as single disks. 100% redundancy of
data means no rebuild is necessary in case of a disk failure, just
a copy to the replacement disk. Transfer rate per block is equal
to that of a single disk. Under certain circumstances, RAID 1 can
sustain multiple simultaneous drive failures. Simplest RAID
storage subsystem design.
Disadvantages: Highest disk overhead of all RAID
types (100%) - inefficient. Typically the RAID function is done by
system software, loading the CPU/Server and possibly degrading
throughput at high activity levels. Hardware implementation is
strongly recommended. May not support hot swap of failed disk when
implemented in "software".
Recommended Applications: Accounting;
Payroll; Financial; Any application requiring very high
availability.
[ TOP ]
RAID 0+1
RAID 0+1: High Data Transfer
Performance
RAID Level 0+1 requires a minimum of 4 drives to
implement.
RAID Level 0+1 is a striping and mirroring
combination without parity. RAID 0+1 has fast data access (like
RAID 0), and single-drive fault tolerance (like RAID 1). RAID 0+1
still requires twice the number of disks (like RAID 1).
Advantages: RAID 0+1 is implemented as a mirrored
array whose segments are RAID 0 arrays. RAID 0+1 has the same
fault tolerance as RAID level 5. RAID 0+1 has the same overhead
for fault-tolerance as mirroring alone. High I/O rates are
achieved thanks to multiple stripe segments. Excellent solution
for sites that need high performance but are not concerned with
achieving maximum reliability.
Disadvantages: RAID 0+1 is NOT to be confused
with RAID 10. A single drive failure will cause the whole array to
become, in essence, a RAID Level 0 array. Very expensive / High
overhead. All drives must move in parallel to proper track
lowering sustained performance. Very limited scalability at a very
high inherent cost.
Recommended Applications: Imaging applications; General fileserver.
[ TOP ]
RAID 2 (ECC)
RAID 2: Hamming Code ECC.
Each bit of
data word is written to a data disk drive (4 in this example: 0 to
3). Each data word has its Hamming Code ECC word recorded on the
ECC disks. On Read, the ECC code verifies correct data or corrects
single disk errors.
RAID Level 2 is one of two inherently parallel
mapping and protection techniques defined in the Berkeley paper.
It has not been widely deployed in industry largely because it
requires special disk features. Since disk production volumes
determine cost, it is more economical to use standard disks for
RAID systems.
Advantages: "On the fly" data error correction.
Extremely high data transfer rates possible. The higher the data
transfer rate required, the better the ratio of data disks to ECC
disks. Relatively simple controller design compared to RAID levels
3,4 & 5.
Disadvantages: Very high ratio of ECC disks to
data disks with smaller word sizes - inefficient. Entry level cost
very high - requires very high transfer rate requirement to
justify. Transaction rate is equal to that of a single disk at
best (with spindle synchronization). No commercial implementations
exist / not commercially viable.
[ TOP ]
RAID 3
RAID 3: Parallel transfer with Parity
The data block is subdivided ("striped") and written on the data
disks.
Stripe parity is generated on Writes, recorded on the
parity disk and checked on Reads.
RAID Level 3 requires a
minimum of 3 drives to implement. RAID Level 3 adds redundant information in the
form of parity to a parallel access striped array, permitting
regeneration and rebuilding in the event of a disk failure. One
stripe of parity protects corresponding strip's of data on the
remaining disks. RAID Level 3 provides for high transfer rate and
high availability, at an inherently lower cost than mirroring. Its
transaction performance is poor, however, because all RAID Level 3
array member disks operate in lockstep.
RAID 3 utilizes a striped set of three or more
disks with the parity of the strips (or chunks) comprising each
stripe written to a disk. Note that parity is not required to be
written to the same disk. Furthermore, RAID 3 requires data to be
distributed across all disks in the array in bit or byte-sized
chunks. Assuming that a RAID 3 array has N drives, this ensures
that when data is read, the sum of the data-bandwidth of N - 1
drives is realized. The figure below illustrates an example of a
RAID 3 array comprised of three disks. Disks A, B and C comprise
the striped set with the strips on disk C dedicated to storing the
parity for the strips of the corresponding stripe. For instance,
the strip on disk C marked as P(1A,1B) contains the parity for the
strips 1A and 1B. Similarly the strip on disk C marked as P(2A,2B)
contains the parity for the strips 2A and 2B.
Advantages: Very high Read data transfer rate.
Very high Write data transfer rate. Disk failure has an
insignificant impact on throughput. Low ratio of ECC (Parity)
disks to data disks means high efficiency. RAID 3 ensures that if
one of the disks in the striped set (other than the parity disk)
fails, its contents can be recalculated using the information on
the parity disk and the remaining functioning disks. If the parity
disk itself fails, then the RAID array is not affected in terms of
I/O throughput but it no longer has protection from additional
disk failures. Also, a RAID 3 array can improve the throughput of
read operations by allowing reads to be performed concurrently on
multiple disks in the set.
Disadvantages: Transaction rate equal to that of
a single disk drive at best (if spindles are synchronized). Read
operations can be time-consuming when the array is operating in
degraded mode. Due to the restriction of having to write to all
disks, the amount of actual disk space consumed is always a
multiple of the disks' block size times the number of disks in the
array. This can lead to wastage of space. Controller design is
fairly complex. Very difficult and resource intensive to do as a
"software" RAID.
Recommended Applications: Video Production and
live streaming; Image Editing; Video Editing; Prepress
Applications; Any application requiring high
throughput.
[ TOP ]
RAID 4
RAID 4: Independent Data disks with
Shared Parity disk Each entire block is written onto a data disk.
Parity for same rank blocks is generated on Writes, recorded on
the parity disk and checked on Reads.
RAID Level 4 requires a
minimum of 3 drives to implement.
Like RAID Level 3, RAID Level 4 uses parity
concentrated on a single disk to protect data. Unlike RAID Level
3, however, a RAID Level 4 array's member disks are independently
accessible. Its performance is therefore more suited to
transaction I/O than large file transfers. RAID Level 4 is seldom
implemented without accompanying technology, such as write-back
cache, because the dedicated parity disk represents an inherent
write bottleneck.
Advantages: Very high Read data transaction rate.
Low ratio of ECC (Parity) disks to data disks means high
efficiency. High aggregate Read transfer rate.
Disadvantages: Quite complex controller design.
Worst Write transaction rate and Write aggregate transfer rate.
Difficult and inefficient data rebuild in the event of disk
failure. Block Read transfer rate equal to that of a single
disk.
[ TOP ]
RAID 5
RAID 5: Independent Data disks with
Distributed Parity blocks Each entire data block is written on a
data disk; parity for blocks in the same rank is generated on
Writes, recorded in a distributed location and checked on Reads.
The array capacity is N-1.
RAID Level 5 requires a minimum of
3 drives to implement. By distributing parity across some or all of an
array's member disks, RAID Level 5 reduces (but does not
eliminate) the write bottleneck inherent in RAID Level 4. As with
RAID Level 4, the result is asymmetrical performance, with reads
substantially outperforming writes. To reduce or eliminate this
intrinsic asymmetry, RAID level 5 is often augmented with
techniques such as caching and parallel multiprocessors.
The figure below illustrates an example of a RAID
5 array comprised of three disks - disks A, B and C. For instance,
the strip on disk C marked as P(1A,1B) contains the parity for the
strips 1A and 1B. Similarly the strip on disk A marked as P(2B,2C)
contains the parity for the strips 2B and 2C. RAID 5 ensures that
if one of the disks in the striped set fails, its contents can be
extracted using the information on the remaining functioning
disks. It has a distinct advantage over RAID 4 when writing since
(unlike RAID 4 where the parity data is written to a single drive)
the parity data is distributed across all drives. Also, a RAID 5
array can improve the throughput of read operations by allowing
reads to be performed concurrently on multiple disks in the
set.
Advantages: Highest Read data transaction rate.
Medium Write data transaction rate. Low ratio of ECC (Parity)
disks to data disks means high efficiency. Good aggregate transfer
rate.
Disadvantages: Disk failure has a medium impact
on throughput. Most complex controller design. Difficult to
rebuild in the event of a disk failure (as compared to RAID level
1). Individual block data transfer rate same as single disk.
Recommended Applications: File and Application servers; Database
servers; WWW, E-mail, and News servers; Intranet servers; Most
versatile RAID level.
[ TOP ]
RAID 6
RAID 6: Independent Data disks with
two Independent Distributed Parity schemes.
Advantages: RAID 6 is essentially an extension of
RAID level 5 which allows for additional fault tolerance by using
a second independent distributed parity scheme (two-dimensional
parity). Data is striped on a block level across a set of drives,
just like in RAID 5, and a second set of parity is calculated and
written across all the drives. RAID 6 provides for an extremely
high data fault tolerance and can sustain multiple simultaneous
drive failures. Perfect solution for mission critical
applications.
Disadvantages: Very complex controller design.
Controller overhead to compute parity addresses is extremely high.
Very poor write performance. Requires N+2 drives to implement
because of two-dimensional parity scheme.
[ TOP ]
RAID 7
(Proprietary)
RAID 7: Optimized Asynchrony for High
I/O Rates as well as High Data Transfer Rates.
Architectural Features: All I/O transfers are
asynchronous, independently controlled and cached including host
interface transfers. All Reads and Write are centrally cached via
the high speed x-bus. Dedicated parity drive can be on any
channel. Fully implemented process oriented real time operating
system resident on embedded array control microprocessor. Embedded
real time operating system controlled communications channel. Open
system uses standard SCSI drives, standard PC buses, motherboards
and memory SIMMs. High speed internal cache data transfer bus
(X-bus). Parity generation integrated into cache. Multiple
attached drive devices can be declared hot standbys.
Manageability: SNMP agent allows for remote monitoring and
management.
Advantages: Overall write performance is 25% to
90% better than single spindle performance and 1.5 to 6 times
better than other array levels. Host interfaces are scalable for
connectivity or increased host transfer bandwidth. Small reads in
multi user environment have very high cache hit rate resulting in
near zero access times. Write performance improves with an
increase in the number of drives in the array. Access times
decrease with each increase in the number of actuators in the
array. No extra data transfers required for parity manipulation.
RAID 7 is a registered trademark of Storage Computer
Corporation.
Disadvantages: One vendor proprietary solution.
Extremely high cost per MB. Very short warranty. Not user
serviceable. Power supply must be UPS to prevent loss of cache
data.
[ TOP ]
RAID 10
RAID 10: Very High Reliability
combined with High Performance
RAID Level 10 requires a
minimum of 4 drives to implement.
Advantages: RAID 10 is implemented as a striped
array whose segments are RAID 1 arrays. RAID 10 has the same fault
tolerance as RAID level 1. RAID 10 has the same overhead for
fault-tolerance as mirroring alone. High I/O rates are achieved by
striping RAID 1 segments. Under certain circumstances, RAID 10
array can sustain multiple simultaneous drive failures. Excellent
solution for sites that would have otherwise gone with RAID 1 but
need some additional performance boost.
Disadvantages: Very expensive / High overhead.
All drives must move in parallel to proper track lowering
sustained performance. Very limited scalability at a very high
inherent cost. Recommended Applications? Database server requiring
high performance and fault tolerance.
RAID 10 arrays are typically used in environments
that require uncompromising availability coupled with
exceptionally high throughput for the delivery of data located in
secondary storage. In recent years a number of mutations of RAID
10 have been developed with similar capabilities. This paper
presents one of the popular alternative implementations and
discusses the relative advantages and disadvantages of RAID 10 and
this alternative.
A RAID 10 array is formed using a two-layer
hierarchy of RAID types. At the lowest level of the hierarchy are
a set of RAID 1 sub-arrays i.e., mirrored sets. These RAID 1
sub-arrays in turn are then striped to form a RAID 0 array at the
upper level of the hierarchy. The collective result is a RAID 10
array. The figure below demonstrates a RAID 10 comprised of two
RAID 1 sub-arrays at the lower level of the hierarchy. They are
sub-arrays A (comprised of disks A1 and A2) and B (comprised of
disks B1 and B2). These two sub-arrays in turn are striped using
the strips 1A, 1B, 2A, 2B, 3A, 3B, 4A, 4B to form a RAID 0 at the
upper level of the hierarchy. The result is a RAID 10. Figure 1
illustrates a RAID 10 array, with each disk in the array
participating in exactly one mirrored set, thereby forcing the
number of disks in the array to be even.
Let us now look at some of the salient properties
of RAID 10. Consider a RAID 10 comprised of d disks and N mirrored
sets (i.e., constituent RAID 1 sub-arrays). Since each disk in the
array participates in exactly one mirrored set, d = 2N.
(a) RAID 10 arrays do not require any parity
calculation at any stage of their construction or operation.
(b) RAID 10 arrays are generally deployed in
environments that require a high degree of redundancy. The ability
to survive multiple failures is a fundamental property of RAID 10.
In fact the maximum number of disk failures a RAID 10 array can
withstand is d/2 = N.
What about the number of combinations of failed
disks that a RAID 10 array can sustain? The number of ways in
which k disks can fail is given by NCk ?2k, since there are NCk
ways in which to choose k mirror groups from N possible choices,
and 2 ways in which to choose a disk within each mirror group.
Therefore the total number of combinations of failed disks that a
RAID 10 can support is:
NC1 ?21 + NC2 ?22 + ? + NCN ?2N
= (2 + 1)N - 1
= 3N -
1
Thus, for a 4 drive RAID 10 containing 2 mirrored
sets, the number of combinations in which disks can fail without
the array being rendered inoperable is 32 - 1 = 8. In fact, these
combinations may be enumerated as follows, with each possible set
of failed disks listed within braces. They are: {A1}, {A2}, {B1},
{B2}, {A1, B1}, {A2, B2}, {A1, B2}, and {A2, B1}.
(c) RAID 10 ensures that if a disk in any
constituent mirrored set fails, its contents can be extracted from
the functioning disk in its mirrored set. Thus, when a RAID 10
array has suffered the maximum number of disk failures it is
capable of withstanding, its throughput rate is no worse than that
of a RAID 0 with N disks. In fact, any combination of N contiguous
independent strips can be read concurrently. The term "independent
strip" is used to denote a strip in a collection of strips that is
not a mirror of any other strip within that collection.
(d) A RAID 10 array that is in a nominal state
can improve the throughput of read operations by allowing
concurrent reads to be performed on multiple disks in the array.
For example, if the strips 1A, 1B, 2A, 2B are to be read from the
array given in figure 1, it is clear that all four strips can be
read concurrently from the disks A1, B1, A2 and B2
respectively.
[ TOP ]
RAID 1E
RAID 1E: While RAID 10 has been
traditionally implemented using an even number of disks, some
hybrids can use an odd number of disks as well. Figure 2
illustrates an example of a hybrid RAID 10 array comprised of five
disks; A, B, C, D and E. In this configuration, each strip is
mirrored on an adjacent disk with wrap-around. In fact this scheme
- or a slightly modified version of it - is often referred to as
RAID 1E and was originally proposed by IBM. Let us now investigate
the properties of this scheme.
When the number of disks comprising a RAID 1E is
even, the striping pattern is identical to that of a traditional
RAID 10, with each disk being mirrored by exactly one other unique
disk. Therefore, all the characteristics for a traditional RAID 10
apply to a RAID 1E when the latter has an even number of disks.
However, RAID 1E has some interesting properties when the number
of disks is odd.
(a) Just as in the case of traditional RAID 10,
RAID 1E does not require any parity calculation either. So in this
category, RAID 10 and RAID 1E are equivalent.
(b) The maximum number of disk failures a RAID 1E
array using d disks can withstand is d/2 . When d is odd, this
yields a value that is the equal to that of a traditional RAID 10
while utilizing one additional disk. What about the number of
combinations of disk failures that RAID 1E can support. It turns
out that RAID 1E is very peculiar in this characteristic when d is
odd. Assume for the sake of notational convenience that d/2 = p.
Then the number of ways in which k disks can fail is d?P-1Ck-1,
since there are d ways to choose the first disk and P-1Ck-1 ways
to choose the remaining k-1 disks from p-1 possible choices.
Therefore, the total number of combinations of failed disks that
this scheme can support is:
d?p-1C0 + d?p-1C1 + ... + d?p-1Cp-1
= d ? (p-1C0 + p-1C1 +
? + p-1Cp-1)
= d ? 2p-1
Thus, for a 5 drive RAID 1E, the total number of
combinations in which disks can fail without the array being
rendered inoperable is 5?22-1 = 10. However, this result also
indicates that as the value of d increases, the ratio of the
number of combinations of disk failures supported by RAID 1E using
d disks decreases with respect to conventional RAID 10 using d-1
disks. In fact, for d > 9, RAID 1E yields a lesser number of
combinations! For instance, while a conventional RAID 10 using 10
disks can support 35 - 1 = 242 combinations of disk failures, RAID
1E using 11 disks can support only 11?25-1 = 176 combinations.
Clearly, RAID 10 is a superior choice when tolerance to a larger
number of combinations of disk failures is considered important.
An even more significant implication of this result is the
following. Since a RAID 1E with an even number of disks is
identical to a traditional RAID 10, A RAID 1E with 10 disks can
support more combinations of failures than a RAID 1E with 11
disks. In general, a RAID 1E with 2N disks can support more
combinations of failures that a RAID 1E with 2N + 1 disks, when N
5. In other words, it is always preferable to utilize an even
number of disks for your RAID 1E than an odd number if you desire
a higher tolerance to disk failures. In other words, it is always
preferable to use a traditional RAID 10!
(c) When a RAID 1E array suffers the maximum
number of disk failures it is capable of withstanding, i.e., d/2 ,
the number of contiguous independent strips that can be accessed
concurrently can be less than d/2 . For example, consider the RAID
1E array displayed in figure 2. Assume that disks A and C have
failed. In this scenario, it is clear that the contiguous strips
4, 5 and 6 cannot be read concurrently although three disks remain
operational. Thus the throughput of a RAID 1E with d disks - where
d is odd - may be no higher under specific access patterns than
that of a RAID 10 with d-1 disks when both arrays experience the
maximum number of sustainable disk failures.
(d) Just as in the case of a traditional RAID 10
implementation, RAID 1E in a nominal state can improve the
throughput of read operations by allowing concurrent reads to be
performed on multiple disks in the array. The fact that there are
more disks than there are mirror sets should intuitively suggest
as much.
Conclusion: RAID 1E offers a little more
flexibility in choosing the number of disks that can be used to
constitute an array. The number can be even or odd. However, RAID
10 is far more robust in terms of the number of combinations of
disk failures it can sustain even when using lesser number of
disks. Furthermore, a RAID 10 guarantees a throughput rate that is
always equal to that which is obtainable from the concurrent use
of all its functioning disks. In contrast, specific access
patterns may not lend themselves to the concurrent use of all
functioning disks under RAID 1E. Therefore, if extremely high
availability and throughput are of paramount importance to your
applications, RAID 10 should be the configuration of
choice.
[ TOP ]
RAID 50 (same as
RAID 05)
RAID 50 array is formed using a
two-layer hierarchy of RAID types. At the lowest level of the
hierarchy is a set of RAID 5 arrays. These RAID 5 arrays in turn
are then striped to form a RAID 0 array at the upper level of the
hierarchy. The collective result is a RAID 50 array. The figure
below demonstrates a RAID 50 comprised of two RAID 5 arrays at the
lower level of the hierarchy ? arrays X and Y. These two arrays in
turn are striped using 4 stripes (comprised of the strips 1X, 1Y,
2X, 2Y, etc.) to form a RAID 0 at the upper level of the
hierarchy. The result is a RAID 50.
Advantage: RAID 50 ensures that if one of the
disks in any parity group fails, its contents can be extracted
using the information on the remaining functioning disks in its
parity group. Thus it offers better data redundancy than the
simple RAID types, i.e., RAID 1, 3, and 5. Also, a RAID 50 array
can improve the throughput of read operations by allowing reads to
be performed concurrently on multiple disks in the set.
[ TOP ]
RAID 53
RAID 53: High I/O Rates and Data
Transfer Performance
RAID Level 53 requires a minimum of 5
drives to implement.
Advantages: RAID 53 should really be called "RAID
03" because it is implemented as a striped (RAID level 0) array
whose segments are RAID 3 arrays. RAID 53 has the same fault
tolerance as RAID 3 as well as the same fault tolerance overhead.
High data transfer rates are achieved thanks to its RAID 3 array
segments. High I/O rates for small requests are achieved thanks to
its RAID 0 striping. Maybe a good solution for sites who would
have otherwise gone with RAID 3 but need some additional
performance boost.
Disadvantages: Very expensive to implement. All
disk spindles must be synchronized, which limits the choice of
drives. Byte striping results in poor utilization of formatted
capacity.
[ TOP ] |
