Download presentation
Presentation is loading. Please wait.
Published byMeredith Collins Modified over 9 years ago
2
N. Xiong@ GSU Slide 1 Chapter 02 Cloud Computing Systems N. Xiong Georgia State University
3
N. Xiong@ GSU Slide 2 Chapter 02 Review and Introduction
4
N. Xiong@ GSU Slide 3 Chapter 02 Scalable Computing Towards Massive Parallelism Enabling Technologies for Distributed Computing Distributed Computing System Models Performance, Security, and Energy- Efficiency References and Homework Problems Chapter 02 Main Contents
5
N. Xiong@ GSU Slide 4 Chapter 02 Levels of Parallelism Bit-level parallelism (BLP) instruction-level parallelism (ILP) Data-level parallelism (DLP) task-level parallelism (TLP) job-level parallelism (JLP) Scalable Computing Towards Massive Parallelism
6
N. Xiong@ GSU Slide 5 Chapter 02 Key issues of the age of Internet computing Efficiency measured in building blocks and execution model to exploit massive parallelism as in HPC. This may include data access and storage model for HTC and energy efficiency. Dependability in terms of reliability and self-management from the chip to system and application levels. The purpose is to provide high-throughput service with QoS assurance even under failure conditions. Adaptation in programming model which can support billions of job requests over massive datasets, virtualized cloud resources, and flexible application service model. Scalable Computing Towards Massive Parallelism
7
N. Xiong@ GSU Slide 6 Chapter 02 The Platform Evolution
8
N. Xiong@ GSU Slide 7 Chapter 02 Distributed Computing Families peer-to-peer (P2P) Grid computing Cloud computing
9
N. Xiong@ GSU Slide 8 Chapter 02 The Top-500 supercomputer performance
10
N. Xiong@ GSU Slide 9 Chapter 02 Architectural evolution of the Top-500 supercomputers
11
N. Xiong@ GSU Slide 10 Chapter 02 Top Five Supercomputers
12
N. Xiong@ GSU Slide 11 Chapter 02 Killer Applications of HPC and HTC Systems
13
N. Xiong@ GSU Slide 12 Chapter 02 Enabling Technologies for Distributed Parallelism Network technologies for distributed computing Software technologies for distributed computing Hardware technologies for distributed computing
14
N. Xiong@ GSU Slide 13 Chapter 02 System Components and Wide-Area Networking
15
N. Xiong@ GSU Slide 14 Chapter 02 Multicore Architecture
16
N. Xiong@ GSU Slide 15 Chapter 02 Memory, SSD, and Disk Arrays
17
N. Xiong@ GSU Slide 16 Chapter 02 Virtual Machines and Virtualization Middleware
18
N. Xiong@ GSU Slide 17 Chapter 02 Virtualization Operations
19
N. Xiong@ GSU Slide 18 Chapter 02 Trends in Distributed Operating Systems Three approaches build a network OS over a large number of heterogeneous OS platforms develop middleware to offer limited degree of resource sharing develop a distributed OS to achieve higher use or system transparency
20
N. Xiong@ GSU Slide 19 Chapter 02 Amoeba vs. DCE
21
N. Xiong@ GSU Slide 20 Chapter 02 Parallel and Distributed Programming Environments
22
N. Xiong@ GSU Slide 21 Chapter 02 Grid Standards and Toolkits for scientific and Engineering Applications
23
N. Xiong@ GSU Slide 22 Chapter 02 Distributed Computing System Models A large number of autonomous computer nodes Interconnected by system-area networks (SAN), local-are networks (LAN), or wide-area networks (WAN) A hierarchical manner
24
N. Xiong@ GSU Slide 23 Chapter 02 System Classification
25
N. Xiong@ GSU Slide 24 Chapter 02 New Challenges new network-efficient processors Scalable memory and storage schemes distributed OS middleware for machine virtualization new programming model effective resource management application program development
26
N. Xiong@ GSU Slide 25 Chapter 02 Cluster Architecture
27
N. Xiong@ GSU Slide 26 Chapter 02 Cluster Design Issues
28
N. Xiong@ GSU Slide 27 Chapter 02 Grid Computing Infrastructures
29
N. Xiong@ GSU Slide 28 Chapter 02 Grid Families
30
N. Xiong@ GSU Slide 29 Chapter 02 Service-Oriented Architectures (SOA) Three kernel standards: Web Service Description Language (WSDL) Simple Object Access Protocol (SOAP) Universal Description Discovery and Integration (UDDI)
31
N. Xiong@ GSU Slide 30 Chapter 02 Service-Oriented Architectures (SOA) General layered architecture for distributed entitiesLayered architecture for web services and grids
32
N. Xiong@ GSU Slide 31 Chapter 02 Integrating Several Entities Together
33
N. Xiong@ GSU Slide 32 Chapter 02 P2P Networks
34
N. Xiong@ GSU Slide 33 Chapter 02 Peer-to-Peer Network Families
35
N. Xiong@ GSU Slide 34 Chapter 02 P2P Computing Challenges Data locality, network proximity, and interoperability routing efficiency and self-organization Fault Tolerance, failure management, and load balancing Security, privacy, and copyright violations
36
N. Xiong@ GSU Slide 35 Chapter 02 Virtualized Cloud Computing Infrastructure
37
N. Xiong@ GSU Slide 36 Chapter 02 Internet Clouds Cloud platform offers a scalable computing paradigm built around the datacenters. Cloud resources are dynamically provisioned by datacenters upon user demand. Cloud system provides computing power, storage space, and flexible platforms for upgraded web-scale application services. Cloud computing relies heavily on the virtualization of all sorts of resources. Cloud computing defines a new paradigm for collective computing, data consumption and delivery of information services over the Internet. Clouds stress the cost of ownership reduction in mega datacenters.
38
N. Xiong@ GSU Slide 37 Chapter 02 Basic Cloud Models
39
N. Xiong@ GSU Slide 38 Chapter 02 Representative Cloud Providers
40
N. Xiong@ GSU Slide 39 Chapter 02 Performance, Security, and Energy-Efficiency System Performance and Scalability Analysis System Availability and Application Flexibility Security Threats and Defense Technologies Energy-Efficiency in Distributed Computing
41
N. Xiong@ GSU Slide 40 Chapter 02 Performance Metrics CPU speed in MIPS network bandwidth in Mbps Job response time network latency quality of service (QoS)
42
N. Xiong@ GSU Slide 41 Chapter 02 Dimensions of Scalability Size Scalability Software Scalability Application scalability Technology Scalability
43
N. Xiong@ GSU Slide 42 Chapter 02 Scalability vs. OS Image Count
44
N. Xiong@ GSU Slide 43 Chapter 02 Amdahl’s Law and Some Improvement
45
N. Xiong@ GSU Slide 44 Chapter 02 System Availability A system is highly available if long mean time to failure (MTTF) short mean time to repair (MTTR) System Availability = MTTF / ( MTTF + MTTR )
46
N. Xiong@ GSU Slide 45 Chapter 02 System Availability
47
N. Xiong@ GSU Slide 46 Chapter 02 Security Threats and Defense Technologies Threats To Systems and Networks Security Responsibilities System Defense Technologies Copyright Protection Data Protection Infrastructure
48
N. Xiong@ GSU Slide 47 Chapter 02 Threats To Systems and Networks
49
N. Xiong@ GSU Slide 48 Chapter 02 Security Responsibilities
50
N. Xiong@ GSU Slide 49 Chapter 02 System Defense Technologies Three generations prevent or avoid intrusions, such as access control policies or tokens detects intrusions timely to exercise remedial actions, like firewalls, Intrusion Detection Systems (IDS), PKI service intelligent responses
51
N. Xiong@ GSU Slide 50 Chapter 02 Energy-Efficiency in Distributed Computing Energy consumption of unused servers Reducing energy in active servers Application layer Middleware layer Resource layer Network layer
52
N. Xiong@ GSU Slide 51 Chapter 02 Energy Consumption of Unused Servers 15% of the full-time servers in a company is idling in the world, around 4.7 million servers are not doing any useful work globally $3.8 billion in energy costs alone and $24.7 billion in the total cost of running non- productive servers
53
N. Xiong@ GSU Slide 52 Chapter 02 Reducing Energy in Active Servers
54
N. Xiong@ GSU Slide 53 Chapter 02 Application layer Challenge: how to design sophisticated multilevel and multi- domain energy management applications without hurting performance First step Find out the relationship between performance and energy consumption
55
N. Xiong@ GSU Slide 54 Chapter 02 Middleware layer A bridge between the application layer and the resource layer Susceptible for applying energy-efficient techniques particularly in task scheduling Need a new cost function covering both makespan and energy consumption
56
N. Xiong@ GSU Slide 55 Chapter 02 Resource Layer Resource include: computing nodes storage units Some approaches: Dynamic power management (DPM) dynamic voltage-frequency scaling (DVFS)
57
N. Xiong@ GSU Slide 56 Chapter 02 Resource Layer
58
N. Xiong@ GSU Slide 57 Chapter 02 Network Layer Two major challenges: The models should represent the networks comprehensively as they should give a full understanding of interactions between time, space and energy; New energy-efficient routing algorithms need to be developed. New energy-efficient protocols should be developed against network attacks.
59
N. Xiong@ GSU Slide 58 Chapter 02 Some References and Further Reading
60
N. Xiong@ GSU Slide 59 Chapter 02 Homework Problems
61
N. Xiong@ GSU Slide 60 Chapter 02 Homework Problems
Similar presentations
© 2025 SlidePlayer.com. Inc.
All rights reserved.