Computer Concepts 8

1
(No Transcript)
2
Chapter Preview

• After this chapter, you should be able to
• Define enterprise computing and describe its key
  characteristics
• Explain how enterprise systems achieve hardware
  and application integration
• Define high-performance computing and list some
  examples of high-performance computing
  applications
• Explain how the term compute-intensive relates
  to high-performance computing

3
Chapter Preview

• Describe the input, processing, output, and
  storage hardware of a typical enterprise
  computing system
• Explain how supercomputers, servers, and personal
  computers contribute to the architecture of
  high-performance computing systems
• Describe and diagram centralized, distributed,
  tiered, and grid computing architectures
• Discuss how quality of service relates to
  large-scale computer systems

4
Chapter 12 Outline

• Section A
• Large-Scale Computing
• Section B
• Components of Large-Scale Computing Systems
• Section C
• Enterprise and High-Performance Architecture
• Section D
• Quality of Service

5
Chapter 12Beyond Desktop Computing

• Section A Large-Scale Computing

Computer Concepts 8th EditionParsons/Oja
6
What is enterprise computing?

• One or more information systems that share data
  and typically provide information to hundreds or
  thousands of users who may be located in diverse
  geographical locations
• Provide more than one service and process large
  numbers of transactions

Page 636
7
What is enterprise computing?
Page 636
8
Whats the difference between an enterprise
system and an information system?

• Scope
• An information system is dedicated to one set of
  related tasks
• An enterprise computing system encompasses the
  tasks associated with multiple information systems

Page 637
9
What is the scale of an enterprise system?

• Classified as large-scale
• Contain from a few dozen to several thousand
  computers

Page 637
10
Can an enterprise computing system expand to meet
the demands of a growing business?

• Information systems often have limited capacity
  for expansion
• Enterprise computing systems are designed for
  scalability
• Scalability refers to the ability of a computer
  system to shrink or grow as requirements change
• Scaling up means increasing individual machine
  performance
• Scaling out means adding more computers to
  increase the overall size of a system

Page 637
11
Can an enterprise computing system expand to meet
the demands of a growing business?
Page 638
12
What is enterprise system integration?

• The process of connecting two or more information
  systems in a way that allows scalability and data
  sharing
• Enterprise hardware integration
• Enterprise application integration (EAI)

Page 638
13
What is enterprise system integration?
Page 639
14
How can an enterprise integrate applications?

• Database linking
• Application linking
• Data warehousing
• Common virtual system

Page 640
15
What is high-performance computing?

• A branch of computer science that focuses on ways
  to optimize the processing capabilities of
  computers
• Designed to handle computationally complex tasks
• Car design
• Weather forecasting
• Earthquake prediction
• Genetic research

Page 641
16
What is high-performance computing?

• The speed of HPC systems is typically measured in
  FLOPS
• MIPS

Page 642
17
What is compute intensive?

• Problem that requires massive amounts of data to
  be processed using complex mathematical
  calculations
• Molecular calculations
• Atmospheric models
• Cosmological research

Page 642
18
How does high-performance computing differ from
enterprise computing?

• High-performance computing focuses on
  compute-intensive problems
• Enterprise computing distributes its processing
  capabilities over a wide range of processing tasks

Page 643
19
Chapter 12Beyond Desktop Computing

• Section B Components of Large-scale Computing
  Systems

Computer Concepts 8th EditionParsons/Oja
20
What types of hardware can be incorporated into
an enterprise system?

• Can include a diverse array of processing,
  storage, input, and output hardware

Page 644
21
What is a legacy system?

• Computer system that has become outdated due to
  technology advances or changing organizational
  requirements
• May perform adequately even though newer
  technology exists
• Legacy systems present several challenges
• Repairs
• Upgrades

Page 645
22
What types of computers form the core of an
enterprise computing system?

• One or more fast servers
• A blade server is a modular electronic circuit
  board containing one or more processors, and
  possibly some storage
• Configured to perform a single dedicated task

Page 645
23
Do enterprise systems use specialized storage
devices?

• Yes. Enterprise systems may need to store many
  gigabytes or even terabytes of data

Page 646
24
What is RAID?

• A storage system that uses a process called
  striping to break files into smaller blocks,
  which are stored across multiple hard disks
• Faster
• More Resilient

Page 646
25
What is a SAN?

• A network of storage devices and data servers
  designed to function as a node on a wider network
• Storage management services

Page 647
26
What is mirroring?

• The process of creating a real-time mirror copy
  of a storage medium
• Requires at least two hard disks
• Happens in real time

Page 647
27
What is NAS?

• Refers to storage devices designed to be attached
  directly to a network without requiring a server
  for management

Page 648
28
Do enterprise systems require special input
devices?

• Data can be entered by a keyboard or a variety of
  other input devices
• MICR (magnetic ink character recognition)
• OCR (optical character recognition)

Page 648
29
Do enterprise systems require special output
devices?

• An enterprise system can use just about any
  standard output device
• Speed and reliability are priorities
• A line printer prints an entire line of text at a
  time
• A print server organizes print jobs and manages
  them using a print queue
• Web-based output

Page 649
30
Does high-performance computing require a
supercomputer?

• Most HPC applications dont require the
  processing speeds that supercomputers provide
• Standard HPC systems are typically required to
  operate at speeds that approach or exceed 1
  gigaFLOPS

Page 650
31
Whats the alternative to a supercomputer for HPC
systems?

• Technological advances make it possible to
  construct a high-performance computing system
  using off-the-shelf personal computers
• Multiprocessor architecture refers to computers
  that use two or more processors

Page 650
32
What role does parallel processing play in
high-performance computing?

• Parallel processing is the simultaneous use of
  more than one processor to execute a program
• Symmetric multiprocessing (SMP) makes use of a
  single operating system to control multiple
  processors
• Massively parallel processing (MPP) is the
  process of linking multiple processors, each with
  its own bus, memory, and operating system

Page 651
33
What role does parallel processing play in
high-performance computing?
Symmetric Multiprocessing
Page 651
34
What role does parallel processing play in
high-performance computing?
Massively Parallel Processing
Page 652
35
What software is used on a typical enterprise
system?

• Multiplatform availability
• Scalability
• Redundancy

Page 653
36
What role does software play in integrating
legacy applications?

• Middleware is a type of software that acts as an
  intermediary between two other software packages
• SOAP

Page 654
37
Does high-performance computing require special
software?

• HPC systems rarely run commercial software
• Custom software provides two important advantages
  for high-performance computing
• Computational speed
• Specialized applications

Page 655
38
Chapter 12Beyond Desktop Computing

• Section C Enterprise and High-performance
  Architecture

Computer Concepts 8th EditionParsons/Oja
39
What is the simplest architecture for enterprise
and high-performance systems?

• A centralized computing system consists of one
  central computer surrounded by terminals

Page 657
40
What is the simplest architecture for enterprise
and high-performance systems?

• The primary processing device is typically a
  mainframe computer, referred to as a host
  computer
• A terminal is an input/output device with no
  storage capacity
• Terminal emulation software

Page 657
41
What are the advantages and disadvantages of a
centralized computing system?

• Advantages
• The host computer performs all processing
• No middleware required
• Terminal malfunctions have little effect on the
  functionality of the entire system
• Easy administration
• Disadvantages
• If host computer fails, entire system fails
• Scalability limits

Page 658
42
What is a distributed computer system?

• A collection of connected computers in which
  processing, data, and application software are
  dispersed among more than one physical computer

Page 659
43
What are the advantages and disadvantages of
distributed computer systems?
Page 660
44
What is a tier?

• A tiered computer system can have two tiers,
  three tiers, four tiers, or more
• n-tier system
• Each tier consists of one or more computers
• Each tier in a distributed system can interact
  with one or more of the other tiers

Page 661
45
What is a tier?
Page 661
46
What are the advantages of tiered architecture?

• Each tier is independent of the others
• Allows an enterprise to switch software more
  easily

Page 661
47
What is a grid computing system?

• A network of diverse computers in which each
  computer is treated as a generic and equal
  resource
• Grid systems require a connection to one or more
  computers running grid management software
• Each computer on the grid runs grid client
  software
• SETI_at_home project

Page 662
48
What is a grid computing system?
Page 662
49
What kinds of problems are best solved using grid
systems?

• Problems that can be divided into smaller
  problems, which can be processed in any sequence
  at any time
• Cracking encryption algorithms
• Analyzing radio telescope signals
• Helping doctors identify cancerous cells
• Predicting earthquakes
• Creating animated movies

Page 664
50
Is the Internet a grid system?

• Some parts of the Internet provide resources for
  grid computing systems
• The Internet and all of the computers it connects
  are better described as a distributed system

Page 665
51
What is clustering?

• A group of two or more devices connected together
  to distribute processing, input, output, or
  storage workloads
• Requires special cluster management software
• Each computer in a cluster is called a cluster
  node

Page 665
52
What is clustering?
Page 665
53
What is fault tolerance?

• A computer systems ability to react gracefully
  to unexpected software or hardware failures
• Cluster failover

Page 666
54
How do clusters optimize resources?

• By allowing cluster nodes to share workloads and
  data
• Load balancing refers to the practice of
  distributing processing and storage tasks among
  the nodes of a cluster in a way that optimizes
  the performance of the entire system

Page 666
55
Is every node in a cluster active at all times?

• A cluster can be configured as an active-active
  cluster so that all nodes are active at the same
  time
• An active-passive cluster contains one node that
  is active at all times, and additional nodes that
  are ready in case of failover

Page 666
56
What is a Beowulf cluster?

• A collection of off-the-shelf computers that is
  interconnected and configured as a cluster to
  handle high-performance computing tasks
• Computers are dedicated to the cluster and run
  cluster applications
• Popular with educational research facilities

Page 667
57
Chapter 12Beyond Desktop Computing

• Section D Quality of Service

Computer Concepts 8th EditionParsons/Oja
58
What is quality of service?

• The level of performance that is provided by a
  computer system
• Reliability
• Availability
• Serviceability

Page 669
59
Who monitors quality of service?

• Usually monitored by the IS Department of an
  organization
• As organizations grow, some outsource QoS
  responsibility

Page 670
60
What are quality-of-service metrics?

• Techniques for measuring a particular
  quality-of-service characteristic

Page 670
61
What are quality-of-service metrics?
Page 671
62
What is risk management?

• The process of identifying potential threats to
  computer equipment and data
• Implementing plans to avoid as many threats as
  possible
• Developing steps to recover from unavoidable
  disasters
• The goals of risk management are to reduce
  downtime, maintain good quality of service, and
  promote business continuity

Page 671
63
What are the most common threats to large-scale
computer systems?

• Natural disasters
• Power outages
• Hardware breakdowns
• Mean time between failures
• Human error
• Software failures
• Security breaches
• Acts of war
• Viruses

Page 672
64
How are large-scale computer systems protected
from threats?

• Deterrents
• Preventative measures
• Corrective procedures
• Detection activities

Page 673
65
Does data center design help to minimize risks?

• A data center is a specialized facility designed
  to house and protect computer systems and data
• A data center typically includes special features
  that enhance security and proactively reduce the
  risk of data loss

Page 674
66
What if a disaster strikes?

• Design a disaster recovery plan
• A disaster recovery plan is a step-by-step plan
  that describes the methods used to secure data
  against disaster, and explains how an
  organization will recover lost data if and when a
  disaster occurs

Page 675
67
What if a disaster strikes?
Page 675
68
What if a disaster strikes?

• An enterprise-wide disaster recovery plan should
• Ensure the safety of people on the premises at
  the time of the disaster
• Continue critical business operations
• Minimize the duration of a serious disruption of
  operations
• Minimize the immediate damage and prevent
  additional losses
• Establish management succession and emergency
  powers
• Facilitate effective coordination of recovery
  tasks

Page 675
69
Conclusion

• You should now be able to
• Define enterprise computing and describe its key
  characteristics
• Explain how enterprise systems achieve hardware
  and application integration
• Define high-performance computing and list some
  examples of high-performance computing
  applications
• Explain how the term compute-intensive relates
  to high-performance computing

70
Conclusion

• Describe the input, processing, output, and
  storage hardware of a typical enterprise
  computing system
• Explain how supercomputers, servers, and personal
  computers contribute to the architecture of
  high-performance computing systems
• Describe and diagram centralized, distributed,
  tiered, and grid computing architectures
• Discuss how quality of service relates to
  large-scale computer systems