raid
TRANSCRIPT
RAIDRAIDRedundant Array of Redundant Array of Independent DisksIndependent Disks
Outli
neOu
tline
What is RAID?What is RAID? RAID configurations used.
RAID configurations used. Performance of each
Performance of each configuration.configuration. Implementations.
Implementations.Way forward with RAID.
Way forward with RAID.
Wha
t is R
AID?
Wha
t is R
AID?
Stands for Redundant Array of
Stands for Redundant Array of
Independent Disks.
Independent Disks.It’s a technology that enables
It’s a technology that enables
greater levels of performance,
greater levels of performance,
reliability and/or large volumes
reliability and/or large volumes
when dealing with data.
when dealing with data.How??How?? By concurrent use of two
By concurrent use of two
or more ‘or more ‘hard disk drives’.
hard disk drives’.How Exactly??
How Exactly?? Mirroring, Mirroring,
Stripping (of data) and Error
Stripping (of data) and Error
correction techniques combined
correction techniques combined
with multiple disk arrays give you
with multiple disk arrays give you
the reliability and performance.
the reliability and performance.
Mean Time Between Failures Mean Time Between Failures (MTBF)(MTBF)
RAID
flav
ors
RAID
flav
ors
Commonly used ones:
Commonly used ones:1.1.RAID 0
RAID 02.2.RAID 1RAID 13.3.RAID 5
RAID 54.4.RAID 10RAID 10Other types used…but
Other types used…but rarely: RAID 2,3,4,6,50……
rarely: RAID 2,3,4,6,50……
RAID 0RAID 0a.a. It splits data among two or more disks.It splits data among two or more disks.
b.b. Provides good performance.Provides good performance.
c.c. Lack of data redundancy means there is Lack of data redundancy means there is no fail over support with this no fail over support with this
configuration.configuration.
d.d. In the diagram to the right, the odd In the diagram to the right, the odd blocks are written to disk 0 and the even blocks are written to disk 0 and the even
blocks to disk 1 such that A1, A2, A3, blocks to disk 1 such that A1, A2, A3, A4, … would be the order of blocks read A4, … would be the order of blocks read
if read sequentially from the beginning.if read sequentially from the beginning.
e.e. Used in read only NFS systems and Used in read only NFS systems and gaming systems.gaming systems.
RAID 0 analysisRAID 0 analysisFailure Rate:Failure Rate: MTBF of RAID0 is roughly proportional to the number MTBF of RAID0 is roughly proportional to the number
of disks in the array.of disks in the array. Pr(disk fail) = 5%, then Pr(disk fail) = 5%, then
Pr(atleastonefails) = 1 – Pr(nonefails) = 1 – [1-0.05]Pr(atleastonefails) = 1 – Pr(nonefails) = 1 – [1-0.05]22 = = 9.75%9.75%
Performance:Performance: The fragments are written to their respective disks The fragments are written to their respective disks
simultaneously on the same sector.simultaneously on the same sector. This allows smaller sections of the entire chunk of data to This allows smaller sections of the entire chunk of data to
be read off the drive in parallel, hence good performance.be read off the drive in parallel, hence good performance.
RAID 1RAID 1•RAID1 is ‘data mirroring’.RAID1 is ‘data mirroring’.
•Two copies of the data are held on Two copies of the data are held on two physical disks, and the data is two physical disks, and the data is
always identical.always identical.
• Twice as many disks are required Twice as many disks are required to store the same data when to store the same data when
compared to RAID 0.compared to RAID 0.
•Array continues to operate so long Array continues to operate so long as at least one drive is functioning.as at least one drive is functioning.
RAID
1 a
nalys
isRA
ID 1
ana
lysis
Failure Rate:Failure Rate: If Pr(disk fail) = 5%, then the
If Pr(disk fail) = 5%, then the
probability of both the drives
probability of both the drives
failing in a 2 disk array is
failing in a 2 disk array is P(both fail) = (0.05)
P(both fail) = (0.05) 2 2 = 0.25%. = 0.25%.
Performance:Performance: If we use independent disk
If we use independent disk controllers for each disk, then
controllers for each disk, then
we can increase the read or
we can increase the read or write speeds by doing
write speeds by doing operations in parallel.
operations in parallel.
RAID 5RAID 5• RAID 5 is an ideal combination RAID 5 is an ideal combination
of good performance, good fault of good performance, good fault tolerance and high capacity and tolerance and high capacity and
storage efficiency.storage efficiency.
• An arrangement of parity and An arrangement of parity and CRC to help rebuilding drive data CRC to help rebuilding drive data
in case of disk failures.in case of disk failures.
• “ “Distributed Parity” is the key Distributed Parity” is the key word here.word here.
RAID
5 a
nalys
isRA
ID 5
ana
lysis
MTBF is slightly better than RAID 0.
MTBF is slightly better than RAID 0.
This is because failure of one disk is
This is because failure of one disk is
not quite a harm. We need more time
not quite a harm. We need more time
if 2 or more disks fail.
if 2 or more disks fail. Performance is also as good as RAID
Performance is also as good as RAID
0, if not better. We can read and write
0, if not better. We can read and write
parallel blocks of data.
parallel blocks of data.One of the drawbacks is that the write
One of the drawbacks is that the write
involves heavy parity calculations by
involves heavy parity calculations by
the RAID controller. Write operations
the RAID controller. Write operations
are slower compared to RAID 0.
are slower compared to RAID 0.
Pretty useful for general purpose uses
Pretty useful for general purpose uses
where ‘read’s’ are more frequent the
where ‘read’s’ are more frequent the
‘write’s’. ‘write’s’.
RAID 10RAID 10a.a. Combines RAID 1 and RAID Combines RAID 1 and RAID
0.0.
b.b. Which means having the Which means having the pleasure of both - good pleasure of both - good
performance and good failover performance and good failover handling.handling.
c.c. Also called ‘Nested RAID’.Also called ‘Nested RAID’.
Impl
emen
tatio
nsIm
plem
enta
tions
Software based RAID:
Software based RAID: Software implementations are
Software implementations are
provided by many Operating
provided by many Operating
Systems. Systems. A software layer sits above the disk
A software layer sits above the disk
device drivers and provides an
device drivers and provides an
abstraction layer between the logical
abstraction layer between the logical
drives(RAIDs) and physical drives.
drives(RAIDs) and physical drives.
Server's processor is used to run the
Server's processor is used to run the
RAID software.
RAID software.Used for simpler configurations like
Used for simpler configurations like
RAID0 and RAID1.
RAID0 and RAID1.
Implementations (contd…)Implementations (contd…)Hardware based RAID:
Hardware based RAID:
A hardware implementation of RAID
A hardware implementation of RAID
requires at least a special-purpose
requires at least a special-purpose
RAID controller.
RAID controller.
On a desktop system this may be built
On a desktop system this may be built
into the motherboard.
into the motherboard.
Processor is not used for RAID
Processor is not used for RAID
calculations as a separate controller
calculations as a separate controller
present.present.A PCI-bus-based, IDE/ATA hard disk
RAIDcontroller, supporting levels 0, 1, and 01.
What’s happening present day?What’s happening present day?RAID 6:RAID 6:
a.a.It is seen as the best way to It is seen as the best way to guarantee data integrity as it uses guarantee data integrity as it uses
double parity.double parity.b.b.Lesser MTBF compared to Lesser MTBF compared to
RAID5.RAID5.c.c.It has a drawback though of It has a drawback though of
longer write time.longer write time.
Road
ahe
ad…
Road
ahe
ad…
The expanded use of RAID-6 and other
The expanded use of RAID-6 and other
dual-parity schemes is a virtual certainty.
dual-parity schemes is a virtual certainty.
RAID vendors to support "fast rebuild"
RAID vendors to support "fast rebuild"
features that can restore hundreds of
features that can restore hundreds of
gigabytes in just an hour or so!!
gigabytes in just an hour or so!!
Striping(of data) would extend across
Striping(of data) would extend across
RAID groups -- not just across drives
RAID groups -- not just across drives
within a group.within a group. Improved disk diagnostic features should
Improved disk diagnostic features should
offer more reliable predictions of
offer more reliable predictions of
impending drive failures, allowing the
impending drive failures, allowing the
rebuild process to begin before an actual
rebuild process to begin before an actual
fault occurs.fault occurs.Hot Spares!!
Hot Spares!!
Refe
renc
esRe
fere
nces
http://searchstorage.techtarget.com/generic/0,295582,sid5_gci1196310,
00.htmlhttp://www.lascon.co.uk/
d008005.htmhttp://en.wikipedia.org/wiki/RAID_0#RAID_0
http://en.wikipedia.org/wiki/RAID Paper - A Case for
Paper - A Case for Redundant Arrays of
Redundant Arrays of Inexpensive Disks
Inexpensive Disks (RAID) - David A
(RAID) - David A Patterson, Garth Gibson,
Patterson, Garth Gibson, and Randy H Katz
and Randy H Katz