multimedia storage and retrieval

8/3/2019 Multimedia Storage and Retrieval

1/52

MULTIMEDIA STORAGE AND

RETRIEVAL


2/52

INTRODUCTION

High performanceCPU,Storage Medium

Max computation rateCPUPerformance,Memory BusBandwidth,Video Bandwidth etc

Isochronous delivery of audioand video objects

Broadly usedMagnetic disks


3/52

MAGNETIC MEDIATECHNOLOGY

Reduction in price per megabyte of high

capacity storage Low cost,High capacity

Multimedia magnetic disk storage

systemsSLEDs(Single Large ExpensiveDisks) and RAIDs(Redundant Array ofInexpensive Disks)


4/52

HARD DISK TECHNOLOGY

Faster mass storage medium

ST506 and MFM Hard drives ESDI Hard drive

IDE

New Enhanced IDE Interface

SCSI

SCSI-2


5/52

ST506 AND MFM Hard Drives

Interfacedefines the signalsand operation of signals b/whard disk controller and hard

disk Simple,Controls platter speed

and movement of heads for adrive

It doesnt define the format ofdata stored on the platter


6/52

How data stored on the platter??Encoding Schemes

MFM

Parallel Data-Series of encoded

pulsesMFM(Modified Frequency Modulation)/FM

Data Separator Circuitryseparates data from sector

info

Drive Capacity--10MB-100MB

RLL(Run Length Limited)Packs 50% more bits thanMFM

Drive Capacity20MB-200MB


7/52

ESDI (Enhanced Small DeviceInterface)-Hard Drive

Improvement over ST506Faster drives

and larger storage capacity

Converts data into serial bit streams anduses RLL for encoding

Defect mapContains locations of bad ordefective sectors on the drive

Supplies cylinder and sector information tocontroller

Data separator circuitry is a part of drive

Drive Capacity80MB-2GB


8/52

IDE(Integrated Device Electronics)

Integrated controller

Parallel data interface-16 bit

Driver Capacity40MB-528MB Supports 2 Drives1 Master and the other Slave

Disadvantage: Jumper settings::add drivephysically removing the drive to

change jumper setting Allows only two drives

Advantage: Both hardware and software interfaces are very simplelow cost

hard disk controller


9/52

New Enhanced IDE Interface

Maximum capacity around 8GB

Supports upto 4 drives

Also supports CD ROM and Tape drives

SCSI(Small Computer SystemInterface)

Used for large volume storage appsinRAID as well as optical disks

Drive Capacity20MB-2GB


10/52

SCSI--2

Its SCSI1 with faster data transfer rates,widerdata paths

Two connector approach

Advantage::

SCSI-2 is back compatible with SCSI-1

Disadvantage::

Wide SCSI-2 has not happened yet


11/52

Why SCSI1 & SCSI2 standardsimportant to multimedia??

Uses single bus for connecting different multimediai/p,o/p devices

Rich commom command set to support all devices

Offers performance upto 10MB/sec with standard 8bittransfers and 40MB/sec in fast and wide mode

Allows definition of vendor-unique command sets tocontrol special devices or special features

The tagged command queuing can improve performance


12/52

Storage Densities and Latencies

Seek latencies::

Overlapped seek:seek on one drive ad then onsecond drive and then reconnect to first drive whenthe seek is complete

Midtransfer seek:device controller can be set to seekduring data transfer via a separate port provided on

SCSI chip

Elevator seek:A track close to the head will be readfirst and then a more distant track although the distanttrack is requested first


13/52


Rotational latencies::

Zero latency read/write:Zero latency readsallow transferring data as soon as the headsettles instead of waiting a disk revolution forthe proper sector

Interleave factor:Keep up with the data streamwithout skipping sectors


14/52


Transfer rate and I/O per second:rate at which data is transferredfrom the drive buffer to the host adapter memory

Max throughput:number of bytes transferred per second

Max Throughput for I/O=Block Transfer Size/Total Latency

Where

Total Latency=T1+T2+T3+T4+T5

T1=seek latencyT2=rotational latency

T3=time required to transfer data from disk to CPUs system memory

T4=firmware latency to setup transfer and complete transaction

T5=Final action on data eg:Display


15/52


I/O per second:

Its a measure of the number of input/outputtransactions performed in a second

I/O per second=Max or overall throughput/block size


16/52


Data Management:

A number of activities are involved in the data

management as

Command queuing:Allows execution of multiple sequentialcommands with system CPU intervention.Its helps inminimising head switching and disk rotational latency

Scatter gather:Scatter is a process whereby data is set forbest fit in the available block of memory or disk.Gatherreassembles data into contiguous blocks on disk or inmemory


17/52

Seek ,Rotational and Data transferRelationships

Data transfer

Rotational latency

Seek latency

I/O % Totalaccesstime

Data block size increasing


18/52

Disk Spanning

Method of attaching multiple drives to a single hostadapter

All drives appear as a single contiguous logical unit

Order of data writingFirst drive first until its full then to

2nd and so on

Good way of increasing storage capacity by adding

incremental drives

Doesnt offer fault tolerance or reliability(goes down) dueto combined MTBF

MTBF=MTBF of single drive/Total no of drives


19/52

RAID (redundant array ofindependent disks)

Redundant array of inexpensive disks

Multiple disk database design

Not a hierarchy

7 levels (6 levels in common use) Set of physical disk drives viewed by the OS as

a single logical drive

Data are distributed across the physical drives

of an array Redundant disk capacity is used to store parity

information => data recoverability

Improve access time and improve reliability


20/52

RAID Level 0Disk striping

Not a true member of the RAID family - does not includeredundancy to improve performance.

User and system data distributed across all disks in thearray in strips.

Imagine a large logical disk containing ALL data. This is

divided into strips (physical blocks or sectors) that aremapped round robin to the strips in the array.

A set of logically consecutive strips that maps exactly onestrip to each array member is referred to as a stripe.

+ If two different I/O requests are pending for two different

blocks of data then there is a good chance that the datawill beon different disks and can be serviced in parallel.

+ If a single I/O request is for multiple logically continuousstrips up to n strips can be handled in parallel.


21/52

RAID Level 0

This software may execute either in the disk sub system or in a hostcomputer


22/52

RAID Level 1Disk mirroring

Redundancy achieved through duplicating all data.

Data stripping is similar to RAID level 0.

Each logic strip is mapped to two physical disks.

Read request can be serviced from either of available 2 disks,which ever involves the minimum seek time and rotational latency

Write request requires both disks to be updated but this can bedone in parallel. (Slower write dictates overall speed).

Recover from failure is simple! (data may still be accessed fromthe second drive

Disadvantage:

Cost requires twice the disk space

Configuration is limited, so used only for system software andother highly critical files.

Improvement occurs if the application can split each read request sothat both disk members participate


23/52

RAID Level 1


24/52

RAID Level 2Bit interleaving ofdata

Utilizes parallel access techniques - All disks participate in the executionof every I/O request.

Spindles of individual drives are synchronized so that each disk head is inthe same position on each disk at any given time.

Data striping very small strips (single byte or word).

Error correcting code is calculated across corresponding bits on eachdisk, and the code bits are stored in corresponding bit positions onmultiple parity disks.

For Hamming Code number of parity (redundant) disks is proportionateto the log of the number of data disks.

On a single read, all disks are simultaneously accessed. The requested

data and the associated error correcting code are delivered to the arraycontroller. Array controller can detect and fix single bit errors.

For write all disks must be accessed.

Good choice only for an environment in which many errors occurtherefore not used much (given high reliability if individual disks and diskdrives).


25/52

RAID Level 2


26/52

RAID Level 3Parity drives

Similar to RAID 2 parallel access with data distributed in small strips.Requires only a single redundant disk because it uses a single parity bit forthe set of individual bits in the same position on all of the data disks.

If drives X0-X3 contain data, and X4 contains parity bits.X4(i) = X3(i) X2(i) X1(i) X0(i)

Redundancy in the case of disk failure, the data can be reconstructed.

If drive X1 fails it can be reconstructed as:X1(i) = X4(i) X3(i) X2(i) X0(i)

Performance can achieve high transfer rates, but only one I/O request canbe executed at one time. (Better for large data transfers in non transaction-oriented environments).


27/52

RAID Level 4sector interleaving withdedicated parity drives

Each disk operates independently - Separate I/O requests satisfied in parallel.Suitable for applications with high I/O request rates and NOT well suited for thoserequiring high data transfer rates.Data striping. (Strips are larger than in lower RAIDs).Bit-by-bit parity strip is calculated across corresponding strips on each data disk, andstored in corresponding strip on the parity disk.Performance write penalty when I/O request is small size. Write must update userdata + corresponding parity bits.

X4(i) = X3(i) X2(i) X1(i) X0(i)If X1(i) is changed to X1(i)

X4(i) = X3(i) X2(i) X1(i) X0(i) = X4(i) X1(i) X1(i)To calculate new parity, the old user data,new user data and old parity strips must beread. Then it can update these two strips with the new data and the newly calculatedparity. Thus each strip write involves two reads and two writes.


28/52

RAID Level 5Block interleavingof data

Same as RAID 4 but parity strips are distributed acrossall disks.Typical allocation uses round-robin.For an n-disk array, the parity strip is on a different disk forthe first n strips and the pattern then repeats.

Avoids potential bottleneck found in RAID 4.


29/52

RAID Level 6

Two different parity calculations are carried out and stored in separateblocks on different disks.Example: XOR and an independent data check algorithm =>makes it possible to regenerate data even if two disks containinguser data fail.

No. of disks required = N + 2 (where N = number of disks required fordata).

Provides HIGH data availability.Incurs substantial write penalty as each write affects two parity blocks.


30/52

COMPARISON OF RAID LEVELS


31/52

COMPARISON OF RAID LEVELS


32/52


33/52


34/52


35/52


36/52


37/52


38/52


39/52


40/52


41/52


42/52


43/52


44/52


45/52


46/52


47/52


48/52


49/52


50/52


51/52


52/52

multimedia storage and retrieval

Documents