17-computersystem-configurationandmethods-091127215714-phpapp01

7/28/2019 17-computersystem-configurationandmethods-091127215714-phpapp01

1/73

Part 2 Computer Systems

System configuration and methods

(Text No. 1 Chapter 5)


2/73


3/73

System Classification and Configuration Technology

Client / Server System Protocols used in networking environment

FTP (File Transfer Protocol) Transfers files to and from a computer running an FTP server

service

Compatible with computers running on different platforms

NFS (Network File System) Service for distributed computing system which provides a

distributed file system, eliminating the need for keeping multiplecopies of files on separate machines

RPC (Remote Procedure Call)

A message passing facility that allows a distributed application tocall services available on various computer on a network. Usedduring remote administration of computers

Web (TCP/IP)


4/73

System Classification and Configuration Technology

Client / Server System

Server Types Print server

Stores spooled print job before sending to printer attached

File server

Store shared files which clients can access Database server

Database management with search functions

User Interface Server (Terminal Services Server)

Provides terminal services to client machines Communication server

Connects asynchronous devices to a LAN or a WAN throughnetwork and terminal emulation software.


5/73

System Classification and Configurations

Client Server System ApplicationThree-tier architecture

1. Data

Database is accessed and needed data referenced

2. Function Message or data processing is performed

3. Presentation

Data exchange with users is implemented

Differentiate 2-tier and 3-tier architectures.


6/73

Client/Server Architecture

Two-tier

Server

Client

Three-tier

Data tier

Function tier

Presentation tier

Application DatabaseClient


7/73


8/73

Client/Server System

Strengths

Open standards resulting multiple vendors

Scalable

Support distributed processing


9/73

Client/Server System

Weaknesses

Redundancy

Difficult integrating wide variety of hardware

and software

Missing standard system development

methodologies


10/73

System Classification and Configurations

Client Server System Application Stored procedure

A technique to speed up the client/server system

Stores on the server the instructions that are frequently used by theclient (SQL statements, code modules etc.)

Since the client can execute the instructions stored in the server byjust calling them, the volume of transmission data and transmissionfrequency are reduced.

Likewise, by translating beforehand the instructions stored on theserver side into an executable format, execution efficiency can befurther improved.


11/73

Stored Procedure


12/73

System Configuration

Reliability

Processing Efficiency

Backups

Clusters


13/73

Reliability Configuration

Simplex

Dual

Duplex


14/73


15/73

System Configurations (Reliability) Dual system

Each device is duplicated to compose a system that performs perfectparallel running of two courses

Redundant processing forverification of output

Also called cross-check processing

performed in ratios of 1:10ms or 1:100ms

In the event of failure of any of the devices, the failed system isseparated and processing is continued with the other processingsystem

High cost and high reliability


16/73

System Configuration (Reliability)

Duplex system

Cold standby mode

Standby system notturned on while on

standby

Hot standby mode

Standby system turned

on


17/73

System Configurations (Reliability)


18/73

Efficiency Configuration

Multiprocessor System

Loosely Coupled

Tightly Coupled

Tandem


19/73

System Configuration (Processing Efficiency)

Multiprocessor system

Multiprocessors share one operating system and

auxiliary storage device

Highly efficient Parallel processing under one operating system

Two types of multiprocessor systems

Loosely coupled

Tightly coupled


20/73

Loosely Coupled MP

In the event of failure,

the processor in which

the failure occurredcan be separated and

the operation can be

continued

High system reliability


21/73

Tightly coupled MP

Multiple processors share

the main storage unit

Synchronization and

information transmission

between processors can be

performed at high speed

Complex communication

control programs are notrequired


22/73

System Configuration(Processing Efficiency)

Tandem multiprocessor system

Multiple processors connected in series resulting in loadbalancing

Front-end processor

Placed in front of main processormainly processes client requests

Back-end processor

placed behind main processor to control large databasemainlyprocesses the database


23/73

System Configuration (Back-up)

Configurations to evade/recover from disasters

Mirror site

Files simultaneously updated. Downtime shortened Hot site

Identical system environments are prepared with a backup

system ready to go into operation

Cold site

Hardware ready but not configured with the system

environment (data centre)


24/73

System Configuration (Clusters)

Strategy that uses communication media toconnect multiple computers for use as a singlecomputer

Dedicated Cluster Multiple computers with same OS and architecture are

connected and used as a single computer

Distributed Cluster

Multiple computers of different types are connected Users can access resources from any of these


25/73

System Modes

System processing mode Centralised processing system

Distributed processing system

System usage mode Batch processing system

Online transaction processing system

Real-time control processing system

System operating modeNon-interactive processing systems

Interactive processing systems


26/73

1. System Processing Modes

Centralised processing system Concentrating data and processing in one location

Extremely efficient

Issues: When the data subject to processing increases, switching to a

computer with a processing capacity that is capable of copingwith that increase is required

SPOF at the host

Easy software maintenance and upgrades (but could be costly

and complex to develop)


27/73

System Processing Mode

Distributed Computing Data and/or processing is distributed into each of the computers

and the user can perform processing using all of the systemresources through the network.

No SPOF Load balancing

Resource sharing

Types Mesh

Vertically distributed configuration Horizontally distributed configuration


28/73

Distributed Computing (Mesh)


29/73

Distributed Computing (Vertical)


30/73

Distributed Computing (Horizontal)


31/73

Processing Modes

Centralized processing

Distributed processing

Vertical Horizontal


32/73

2. System Usage Modes 1) Batch Processing

Centralized processing using a dedicated computer

All related data collected and processed together to get the required result.

Payroll calculation

Marking and aggregation of examinations

Statistical analysis

Centre batch processing

Remote batch processing

RJE (Remote job entry)


33/73

Batch Processing


34/73

Centre Batch Processing

Open batch processing

User does everything from data storage to computermanipulation

Closed batch processing

User hands over procedure and data to operator whodoes the computer processing

Cafeteria system

User registers processing procedure and data incomputer and leave the rest of the operation to theoperator


35/73

Operating System for Batch Processing

Job control language (JCL) to implement automatic jobprocessing by specifying

Job name

Storage location of program to be used Storage location of data to be processed

Area of work file and output file

SPOOL

Data subject to processing, as well as the processing results, are

stored at high speed in an auxiliary storage device, which is theonly device with which the processor exchanges data

Frees up processors to perform process-oriented tasks, not I/Otasks


36/73

System Modes

2) Online transaction processing (OTP) system

Systems in which terminal devices at remote locations

and computers are connected through communication

lines

A great number of these systems are online transaction

processing (OLTP) systems, in which the data generated

as a result of a transaction is processed in real time


37/73

OTP

Characteristic

the data and information subject to processing is normally

managed as a centrally controlled database

Some example applications Seat reservation and ticketing system in the transportation business

Deposit and money order systems as well as investment and loan

systems in the finance sector

Stock exchange system in the securities sector Sales inventory management system and customer information

control system in wholesale and retail businesses


38/73

OTP Conditions

Simultaneous execution control (exclusive control) toenable simultaneous response to the requests ofmultiple users

Programs performing OTP must be reentrant programs Can be executed again before their former execution is

completed

Program area and the data area are separated for eachtransaction

Simultaneous processing of multiple requests can be performed

Correct results can be returned for each of these requests

Likewise, since the resources are simultaneously sharedby multiple users, it is necessary to performsimultaneous execution control (exclusive control) ofthe resources


39/73

OTP

Failure recovery

hardware (processor, disk, printer, etc.)

breakdowns, as well as application failures,must be detected and coped with promptly

Robust failure detection and failure recovery

functions

Backup and recovery

Rollback strategy


40/73

System Usage Modes

3) Real-time control processing system

The performance of immediate (real-time)

calculation processing of the informationobtained and the output of the results as controlinformation is called real-time control

The system adopting the real-time approach is

generically known as the real-time processingsystem

Timeliness is of utmost importance


41/73

Real-time control processing system Characteristics

Seldom uses input devices such as a keyboard or output devices such as aprinter

In most cases, the input devices consist of diverse sensors, and the outputdevices of actuators and other control devices

Likewise, the processors, mainly miniaturized microprocessors, are oftencomposed of main storage units that store programs and data

Some application examples

Flight control system

Air-traffic control system

Power supply monitoring system

Industrial robot control

Motor fuel control system and braking system

Household electric appliances such as rice cookers, washing machines andair conditioners


42/73

Real-time control processing system

Real-time operating system Multi-task processing function

Task switching function

Function to minimize the load of the monitor itself Interfaces required

RS-232C (Recommended Standard-232C)

USB (Universal Serial Bus)

Centronics interface SCSI (Small Computer Systems Interface)

GPIB (General Purpose Interface Bus)


43/73

3) System operating modes

Non-interactive processing systems

Example: batch processing

Since processing is performed after theprocedure is indicated, once it has started,

humans cannot intervene in the processing


44/73

System operating modes

Interactive processing systems

Characteristic

humans can provide indications or perform changeswhile interacting with the computer

Functions of the software

GUI

Windows


45/73

System Performance

Measures for computer system performance

evaluation

1. Processing Time2. Processing Efficiency


46/73

1. Processing Time

Time taken to execute data processing by

the computer system

2 standard ways of measurementTurn Around Time (TAT)

Response Time


47/73

TAT

Time taken to return the results when a

batch processing job is submitted to the

computer in batch processing systems In systems with high processing capacity,

TAT is reduced by allocating a high priority

to special jobs


48/73

Response Time

In OTP systems

time to get a produce a response from the computer from the time

the transaction is input

the main aim in the development of online systems is to shortenthe response time, thereby, improving the efficiency of the

computer processing


49/73

2. Processing Efficiency

The ability of the computer system to

process data

ThroughputCommand/Instruction Mix

MIPS

FLOPS


50/73

Throughput

Volume of work that can be processed by

the computer system in a given time

Batch processing : The number of jobs that canbe processed within a given time

Online transaction processing: The number of

transactions that can be processed within a

given time


51/73

Instruction Mix

The combination of the execution time for representative

instructions and frequency of such instruction occurrences

found in programs represent the performance of the

computer's processor Representative programs with the individual instruction

can be divided into two kinds:

Commercial mix: These are frequently used in business processing

and uses mainly transmission instructions Gibson mix: These are frequently used in scientific calculations

and uses mainly the calculation instructions


52/73

Instruction Mix


53/73

MIPS

Million Instructions Per Second

Average number of instructions that a

processor can execute in units ofmillions/sec


54/73

FLOPS

Floating Point Operations Per Second

MIPS : representative measure of business

processing performance evaluation

FLOPS : the number of floating point

calculations possible in one second for

scientific calculations


55/73


56/73

Test Program To simulate the actual business operations under

typical (and non-typical) usage

Benchmark Actual working programs are executed to measure systems

processing efficiency TPC benchmark TPC-A : Banking operations based on the I/O model of the ATM

TPC-B : Database system in a batch processing environment.

TPC-C : Order entry model

TPC-D : Decision support applications

SPEC Distributed processing benchmark especially UNIX systems

SPEC-int : integer type calculations

SPEC-fp : floating point type calculations


57/73

Benchmark

PC Processors WebBenchWeb server

Trade 2eBusiness server using Java

MMB2Mail server

NetBenchFile server


58/73


59/73

System Reliability

One construct of safety and efficiency is RASIS Reliability

Availability

Serviceability Integrity

Security

Includes measures of:

Mean Time Between Failure (MTBF) Mean Time To Repair (MTTR)


60/73

MTBF

Degree of reliability (R in RASIS)

Average time between the occurrence of one failure

to the nextMTBF = Total normal operation time

Total number of times of normal execution


61/73

MTTR

Degree of serviceability (S) in RASIS

Average time in which equipment is not

acting normally

MTTR = Repair Time / n(Failures)


62/73

Availability

Degree of availability (A) is RASIS

Availability = MTBF / (MTBF + MTTR)

However, the measure is operated ondifferently when applied to different system

configurations

Serial systemParallel system


63/73

Serial Systems Availability

Serial system's availability

=

availability of equipment 1 *availability of equipment 2 , ... *

availability of equipment n


64/73

Parallel Systems Availability Availability of serial systems connected in parallel

Duplex system If one processor fails, the remaining one can still operate normally and the system

still operates normally.

The only situation where the entire system is stopped is when both the processorswere to breakdown simultaneously

Availability of a parallel system= 1 - ( ( 1 - availability of equipment 1) *

( 1 - availability of equipment 2) )

This first value of 1 in the formula represents a situation where there is nofailure


65/73

Availability of 2-out-of-3 systems

In a 2-out-of-3 system, the

system can function normally

using only 2-out-of-3

processors. 3rd processor isadded for redundancy

However, the system will fail if

either all the processors failed or

two of the three processors wereto fail.


66/73



67/73


The system will operate for cases 1 to 4

All the equipment has an availability of 0.9 and a failure of 0.1

Availability in case 1 = 0.9 x 0.9 x 0.9 = 0.729

Availability in case 2 = 0.9 x 0.9 x 0.1 = 0.081 Availability in case 3 = 0.9 x 0.1 x 0.9 = 0.081

Availability in case 4 = 0.1 x 0.9 x 0.9 = 0.081

Availability of a 2 out of 3 system = 0.729 + 0.081 + 0.081 + 0.081 =

0.972


68/73

Reliability Design The following factors should be considered:

Fail safe safety aspects in order to minimise the effect on the other parts

when failure occurs

For example, the traffic would automatically switch to red if

the traffic control light system were to go down. This wouldhelp to prevent accidents that may result from the system'sfailure.

Fail soft For example, if a power failure were to occur in a hospital, the

minimum amount of lights would automatically be available

and priority given to life support or life saving equipmentwhen the generators are run.

Fail (fool) proof To prevent the effect of mis-operation by the human element.

For example, input checking is done and re-entry is made tothe misentered data.

R li bilit bj ti d l ti


69/73

Reliability objectives and evaluation

(RASIS)

Reliability This is measured as the MTBF (Mean Time Between

Failures). This can be considered as the normal operation time ofthe system.

Availability This represents the possible usage ratio of the computer

system. This is computed as A = MTBF / ( MTBF + MTTR)

Serviceability This represents the ease of maintenance of the computer system.

This is computed as MTTR (Mean Time To Repair). This can be

considered as the down time for the system. Integrity

This represents the ability to prevent the data from being corrupted

Security This represents the ability to ensure the security of the data


70/73


(Bathtub Curve)

Curve of failure rate against time

Initial failure period (Burn-in) Failure rate for when system is initially installed

Decreasing failure rate for stable operations

Ad hoc failure period (Useful Life) Chance events

Fairly constant failure rate

Systems steady period

Critical failure (Wear-out) Fatigue/aging of system

Increased failure rate

Time to change/modify/upgrade/maintain


71/73

Bathtub Curve


72/73


(Uninterrupted Operation)

There are usage situations which that do not allow

the system to stop operation. This means

continuous operation is required.

life support systems in the hospitals or banking systems

Implemented by using

UPS (Uninterruptible Power Supply)

Multiplexing systems Fault-tolerant systems


73/73

Financial Consideration

Setup cost

Operating cost

17-computersystem-configurationandmethods-091127215714-phpapp01

Documents

Transcript of 17-computersystem-configurationandmethods-091127215714-phpapp01