architecture of intelligent disk subsystem a disk subsystem is a hard disk server. servers are...

www.bookspar.com | VTU NOTES | QUESTION PAPERS | NEWS | VTU RESULTS | FORUM | BOOKSPAR ANDROID APP

Architecture of intelligent Disk subsystem

•A disk subsystem is a hard disk server.

•Servers are connected to the connection port of the subsystem using standard I/O techniques such as SCSI, fibre channel etc.

•The internal structure of the disk subsystem is completely hidden from the server. The server sees on the hard disks.


• The connection ports are extended to the hard disk subsystem by means of internal I/O channels.

• A controller lies between connection ports and hard disks.• The controller increases the data availability and data access

performance with the help of RAID (Redundant Array of Independent Disks)

• Some controllers serve as instant copiers and remote mirroring.

• A cache is also used to accelerate read and write access to the server.


• Small disk subsystems have 1 to 2 connections for servers or storage networks. 6-8 hard disks capable of storing 500 GB of data.

• Larger disk subsystems have more than 10 ports, redundant controllers & several I/O channels. Can store 10 tera bytes of data.


• Most disk subsystems have the feature that they can be connected to different servers. (storage pooling)

• All servers are directly connected to the disk subsystem or indirectly through storage network.

• Here each server can be assigned free storage.


Hard disks and internal I/O channels

• With regard to performance, It is better to use smaller hard disks, so that more hard disks are available in the disk subsystem. At the expense of max capacity.

• Here overall load is spread over more arms and read/write heads and over more I/O channels.

• Standard I/O techniques like SCSI and fibre channel are often used for the internal I/O channels between connection ports and controller and between controller and internal hard disks.


Design of I/O channels: Cases

• Active• In active cabling the individual hard disks are only connected

via one I/O channel. If the access path fails, then it is no longer possible to access the data.

• Active/passive• In active/passive cabling the individual hard disks are

connected via two I/O channels • In normal operation the controller communicates with the

hard disks via the first I/O channel and the second I/O channel is not used.

• In the event of the failure of the first I/O channel, the disk subsystem switches from the first to the second I/O channel.


• Active/active (no load sharing)• In this cabling method the controller uses both I/O channels in

normal operation.• The hard disks are divided into two groups: in normal

operation• the first group is addressed via the first I/O channel and the

second via the second I/O channel.• If one I/O channel fails, both groups are addressed via the

other I/O channel.


• Active/active (load sharing)• In this approach all hard disks are addressed via both I/O

channels in normal operation.• The controller divides the load dynamically between the two

I/O channels so that the available hardware can be optimally utilized.

• If one I/O channel fails, then the communication goes through the other channel only.


active cabling is cheapest and simplest but no protection against failure.

Active/passive is the minimum requirement needed to protect against failure.

Active /active cabling with load sharing best utilizes underlying H/w.


JBOD: Just a Bunch Of Disks

Disk subsystems are classified based on :i. No controllerii. RAID controlleriii. Intelligent controller.if the disk subsystem has no internal controller ,it is just an

enclosure with disks.(JBOD). Here connections for I/O channels and power supply are taken outward at a single point. JBOD is simpler to manage. A server treats these disks as independent disks. If there are 16 disk subsystems,16 device addresses are required.

->JBOD does not support virtualization or RAID


Storage Virtualization Using RAID

• RAID ,initially was called as REDUNDANT ARRAY OF INEXPENSIVE DISKS

• Now its called as REDUNDANT ARRAY OF INDEPENDENT DISKS.

• Disk subsystems that support RAID are called as RAID arrays• RAID has two goals: i) to increase performance by striping ii) to increase fault tolerance by redundancy


RAID 0

1

95

2

106

3

117

0

Host

RAIDControllerRAIDController

RAID 1

Block 1Block 1 Block 1Block 1Block 1Block 1Block 0Block 0Block 0Block 0

Host

Block 0Block 0 RAIDControllerRAIDController


• The bundle of physical hard disks brought together by the RAID controller is known as Virtual disk.

• A server connected to a RAID system only sees the virtual hard disk.

• The fact that RAID controller actually distributes the data over several physical disks is completely hidden to the server.

• A RAID controller can distribute the data that a server writes to the virtual hard disks among the individual physical hard disks in various manners. Theses different procedures are known as RAID levels.


• Hot spare disks: these are not used in normal operation.• If a disk fails the RAID controller immediately begins to copy

the data of the remaining intact disk onto a hot spare disk.• After replacing the defective disk , this is included in the pool

of hot spare disks.


RAID levels

RAID 0: Block by Block striping: RAID 0 distributes the data that the server writes to the virtual hard disk onto one physical hard disk after another block by block.

In the fig there are 4 physical hard disks. The server writes the Blocks A,B,C,D.. Onto the virtual hard disk one after the other .

The RAID controller distributes the sequence of blocks onto the individual physical hard disks (First block A to the first physical disk,B to second disk etc..) after D ,E is written to first disk,block F to second …


• RAID 0 increases the performance of the virtual hard disk as follows: the individual hard disks can exchange data with the RAID controller through I/O channel more quickly . When first block is written into first disk, second is sent into second disk etc..

• RAID 0 is the choice for applications for which the maximum write performance is more important than protection against failure. Examples are the storage of multimedia data for film and video production.

• RAID 0 is used as a fast store for segments in which intermediate results for complex requests are to be temporarily stored.

• Not fault tolerant (no redundancy)


RAID 1: Block by block mirroring

• In RAID 1 fault tolerance is given importance. Here two hard disks are brought together to form a virtual hard disk by mirroring the data on two physical hard disks.

• If the server writes a block to the virtual hard disk, the RAID controller writes this block to two physical hard disks.

• The individual copies are called as mirrors.• The performance increases only in read operation. while

reading load can be divided between two disks. But writing is slow as data has to be sent into disks.

• RAID 1 performance and capacity are limited . So it’s a good choice for storing small databases.


RAID 0+1/RAID 10

• Striping and mirroring combined: RAID 0+1 and RAID10 both represent a two stage virtualization hierarchy.

• Fig 2.11 shows the principle behind RAID 0+1 (mirrored stripes)

• In the example eight physical hard disks are used.• The RAID controller initially brings together each four physical

hard disks to form two virtual hard disks that are only visible within the RAID controller by means of RAID 0 (striping)

• In the second level, it consolidates these two virtual hard disks into a single virtual hard disk by means of RAID 1(mirroring). This virtual hard disk is visible to the server.


RAID 0+1


RAID 10 (striped mirrors)• Here the sequence RAID 0(striping) and RAID 1 (Mirroring)is

reversed in relation to RAID 0+1(mirrored stripes) (fig 2.12)• Here the RAID controller initially brings together the physical

hard disks in pairs by means of RAID 1(mirroring) to form a total of four virtual hard disks that are only visible within the RAID controller.

• In the second stage, the RAID controller consolidates these four virtual hard disks into a virtual hard disk by means of RAID 0 (striping).

• Here also this last virtual disk is visible to the server.


RAID 0+1 vs RAID 10

• When using RAID 0 the failure of a hard disk leads to the loss of the entire virtual hard disk. In RAID 0+1 the failure of a physical hard disk is thus equivalent to the effective failure of four physical hard disks.(fig 2.13). If one of the disk is lost, data is lost. It may be possible to reconstruct the data (but difficult).

• In RAID 10 after the failure of the individual hard disk, additional failure of a further hard disk does not arise. (fig 2.14)


• RAID 10 has higher fault tolerance than RAID 0+1.• The cost involved in restoring the hard disk after a failure is

less in case of RAID 10• In RAID 10 only one physical hard disk has to be recreated

while in RAID 0+1 a virtual hard disk must be recreated that is made up of four physical disks.


RAID 4 and RAID 5 (parity instead of mirroring)

• Even though RAID 10 provides excellent performance ,the problem is mirroring doubles the required storage capacity.

• In RAID 4 and RAID 5 all mirror disks are replaced with a single parity hard disk.(fig 2.15)

• The server writes the blocks A,B,C,D,E etc to the virtual hard disk sequentially. The RAID controller stripes the data blocks over the first four physical hard disks. instead of mirroring the RAID controller calculates a parity block for every four blocks and writes this onto the fifth physical hard disk.

• For ex the controller calculates the parity block• P ABCD for blocks A,B,C and D. If one of the block fails the controller can

reconstruct the data using the three other disks and parity disk.• RAID 4 saves three physical hard disks.• server sees only virtual disk.


RAID Redundancy: Parity

Parity Disk

1

95

3

117

0

0 1 2 34 5 6 7

4

6

1

7

18

Host

RAIDControllerRAIDController

Parity calculation 4 + 6 + 1 + 7 = 18The middle drive fails:

4 + 6 + ? + 7 = 18? = 18 – 4 – 6 – 7? = 1

?


• The parity block is calculated using XOR operation.• P ABCD =A XOR B XOR C XOR D (ex 2.15)• Changing a data block requires changing of value of parity block.ie each write operation to the virtual hard disk requires:i.the physical writing of the data block.ii.The recalculation of the parity blockiii.The physical writing of the newly calculated parity block

Write penalty: The extra cost for write operations in RAID4 and RAID 5 is called the write penalty.


• Ex: fig 2.16 shows a case where server changes block D on the virtual hard disk.

• The controller reads the data block and the parity block form the disk into its cache . then it uses XOR operation to calculate the difference between old and new parity blocks.


• RAID 4 and RAID 5 implementations are capable of reducing the write penalty .ex ,if large data are written sequentially, then the RAID controller can calculate parity blocks from the data flow without reading the old parity block from the disk.

• If the cache size is large it holds frequently changed parity blocks after writing to the disk.

• RAID 4 saves all parity blocks onto a single physical disk. Thus the parity disk becomes the performance bottleneck of RAID if there are high write operations.


• RAID 5 distributes the parity blocks over all hard disks. (fig 2.17)• For ex:P ABCD goes to the 5th disk(H) while P EFGH goes to the disk D.• RAID4 and RAID5 suffer from write penalty. (RAID4 not used in general)• RAID4 and RAID5 can withstand the failure of a physical hard disk. Parity

blocks help to restore.• RAID4 is not used in practice.• Some RAID5 implementations have second parity disk to protect data.


Comparison of RAID levels


RAID 0 is suitable for which the maximum write performance is more important than protection against the failure of a disk. ( multimedia,or physical experiments where large time is required)

RAID 1 Performance and capacity are limited because only two physical hard disks are used . useful for small databases.

RAID 10 is used in situations where high write performance and high fault tolerance is needed. (Used for database log files)

RAID4 and RAID 5 save disk space at the expense of poorer write performance.

architecture of intelligent disk subsystem a disk subsystem is a hard disk server. servers are...

Documents

hard disk subsystem

internal io channels

internal hard disks

hard disk server

design of io channels

individual hard disks

hard disks capable

smaller hard disks