Slideshow:
How to protect data from a disk crash
Disk mirroring: principle
Disk mirroring: write
and read operations
Disk mirroring: additional benefits and
cost
Intro to RAID (Redundant Array of Independent Disks)
RAID level 4 using n disks
Example of RAID level 4
using 4 disks
using even parity
disk1, disk2 and disk3 are data disks
disk4 stores the computed parity on all the data disks
RAID level 4: read operation
RAID level 4: write operation
Example write operation:
updating a disk block on disk2
Example write operation:
updating a disk block on disk2
Example write operation:
updating a disk block on disk2
Example write operation:
updating a disk block on disk2
The parity disk makes recovery possible after a disk failure !!!
Example write operation:
updating a disk block on disk2
Property of RAID disks
Recover from
disk failure using RAID disks
Recover from
disk failure using RAID disks
Problem with the
RAID level 4 algorithm
Solution: RAID 5 - spread parity function
over
multiple disks
Example of RAID 5
- parity cylinder (RAID 5)
- Only way to recover
from a disk crash:
- Redundancy (e.g., backup copy)
- Techniques that deploy
redundancy:
- disk mirroring
- RAID (Redundant Array of Independent Disks)
- disk mirroring
- Mirroring:
- A disk is
a logical unit
- Physically a
disk consists of
two (physical) disks
- A disk is
a logical unit
- Write data to the disk:
- Write data to both disks
- Read data from the disk:
- Read data from any one of disks
- Advantage:
- Read operation can be
twice as fast
- Because you can use one disk to process half of the read requests and use the 2nd disk to process the other half of the read requests
(Write operation is about the same speed)
Disadvantage:
- Very expesnive (you must use twice the number of hard disks !!!)
- Read operation can be
twice as fast
- Comment:
- The RAID technique
has many "levels"
- RAID level 1 = the same as disk mirroring
- The RAID technique
has many "levels"
- RAID level 4:
- Use n disks as follows:
- n − 1 disks are normal disks (i.e., you write data blocks to these disks)
- Disk n is
a parity disk:
- Every sector on disk n stores the result of the parity value of all the (corresponding) sectors on the other n − 1 disks
Example:
- Use n disks as follows:
- Read operation from
a (data) disk:
- A read operation
will access
a "normal" data disk
- The read operation is
processed in the
same way as
without RAID
I.e.:
- No special procedure is necessary to read data in RAID level 4
- A read operation
will access
a "normal" data disk
- Write operation
to a (data) disk:
(you must also
update the parity !!!)
- Write the
data to the
(data) disk
- Compute the
new value of
the parity
- Write the new parity value to the parity disk
- Write the
data to the
(data) disk
- Example: a write operation
that update a data block stored
on disk2
- We first read the
data block
from disk 2:
- Suppose we modify/update the
data block
to 1111:
- We
must first
update the
data disk:
- And then
update the
parity disk with
the correct (= even) parity:
The new value of the parity disk is computed as follows:
new-parity = old-parity XOR ( new-data XOR old-data )
This will ensure that:
parity disk = disk1 XOR disk2 XOR disk3
- We first read the
data block
from disk 2:
- Property of
a RAID disks:
- Consider the following
RAID disk content using
even parity:
- The result of the
XOR operation on
the corresponding sectors
of all n disks is
always equal to
000000000000...000:
- Consider the following
RAID disk content using
even parity:
- How to recover from a
disk failure using
RAID 4:
- Suppose
any one disk fails,
for example,
disk #3:
- We can recover the
content of
disk #3 using the
XOR operation
on sectors from
the other n−1 disks
Example on how to recover one sector:
- Suppose
any one disk fails,
for example,
disk #3:
- Problem with
RAID level 4:
- Each write operation
must
update the
parity disk
Result:
- The parity disk will likely to become a bottleneck !!!
- Each write operation
must
update the
parity disk
- RAID Level 5 protocol:
- Parity cylinder !!!
- Use a finer granularity to
compute the parity:
cylinder
- The parity cylinder is
located on
different disks.
E.g.:
- Cylinder 1: the parity cylinder is located on disk 1
- Cylinder 2: the parity cylinder is located on disk 2
- And so on...
- Use a finer granularity to
compute the parity:
cylinder
Example:
- Parity cylinder !!!