SlideShare ist ein Scribd-Unternehmen logo
1 von 64
Downloaden Sie, um offline zu lesen
HPC Cluster & Cloud Computing




        āļāļīāļ•āļ•āļīāļĢāļąāļāļĐāđŒ āļĄāđˆāļ§āļ‡āļĄāļīāđˆāļ‡āļŠāļļāļ‚
        (Kittirak Moungmingsuk)
        kittirak@clusterkit.co.th
        Feb 9, 2010 @ NIDA
What's ?




           E=mc2




                   2
Mass–energy equivalence




 E : Energy (āļžāļĨāļąāļ‡āļ‡āļēāļ™)
 m : mass (āļĄāļ§āļĨ)
 c : āļ„āļ§āļēāļĄāđ€āļĢāđ‡āļ§āđāļŠāļ‡ (āļĄāļēāļˆāļēāļāļ āļēāļĐāļēāļĨāļ°āļ•āļīāļ™āļ„āļģāļēāļ§āđˆāļē celeritas āđāļ›āļĨāļ§āđˆāļē āļ­āļąāļ•āļĢāļēāđ€āļĢāđ‡āļ§)




                                                                     3
āđ„āļ­āļ™āđŒāļŠāđ„āļ•āļ™āđŒāļ™āļąāļšāļ–āļ·āļ­āļžāļļāļ—āļ˜
  āļ„āļļāļ“āļ„āļīāļ”āļ­āļĒāđˆāļēāļ‡āđ„āļĢ ?




                      4
āđāļ›āļĨāļˆāļēāļāļ•āļ­āļ™āļŦāļ™āļķāđˆāļ‡āđƒāļ™ “The Human Side”
 "āļĻāļēāļŠāļ™āļē āđƒāļ™āļ­āļ™āļēāļ„āļ• āļˆāļ°āļ•āđ‰āļ­āļ‡āđ€āļ›āđ‡āļ™āļĻāļēāļŠāļ™āļēāļŠāļēāļāļĨ āļĻāļēāļŠāļ™āļēāļ™āļąāđ‰āļ™āļ„āļ§āļĢāļ­āļĒāļđāđˆāđ€āļŦāļ™āļ·āļ­
 āļžāļĢāļ°āđ€āļˆāđ‰āļēāļ—āļĩāļĄāļĩāļ•āļąāļ§āļ•āļ™ āđāļĨāļ°āļ„āļ§āļĢāļˆāļ°āđ€āļ§āđ‰āļ™āļ„āļģāļēāļŠāļ­āļ™āđāļšāļšāļŠāļīāļ—āļ˜āļąāļ™āļ•āđŒ (āļ„āļ·āļ­āđ€āļ›āđ‡āļ™āđāļšāļš
           āđˆ
 āļŠāļģāļēāđ€āļĢāđ‡āļˆāļĢāļđāļ›āļ—āļĩāđˆāđƒāļŦāđ‰āđ€āļŠāļ·āđˆāļ­āļ•āļēāļĄāđ€āļžāļĩāļĒāļ‡āļ­āļĒāđˆāļēāļ‡āđ€āļ”āļĩāļĒāļ§) āđāļĨāļ°āđāļšāļšāđ€āļ—āļ§āļ§āļīāļ—āļĒāļē(āļ„āļ·āļ­āļžāļķāļ‡āđ€āļ—āļ§āļ”āļē
                                                              āđˆ
 āđ€āļ›āđ‡āļ™āļŦāļĨāļąāļāđƒāļŦāļāđˆ) āļĻāļēāļŠāļ™āļēāļ™āļąāđ‰āļ™ āđ€āļĄāļ·āđˆāļ­āļ„āļĢāļ­āļšāļ„āļĨāļļāļĄāļ—āļąāđ‰āļ‡āļ˜āļĢāļĢāļĄāļŠāļēāļ•āļīāđāļĨāļ°āļˆāļīāļ•āđƒāļˆ āļˆāļķāļ‡āļ„āļ§āļĢāļĄāļĩ
 āļĢāļēāļāļāļēāļ™āļ­āļĒāļđāđˆāļšāļ™āļŠāļēāļĄāļąāļāļŠāļģāļēāļ™āļķāļāļ—āļēāļ‡āļĻāļēāļŠāļ™āļē āļ—āļĩāđˆāđ€āļāļīāļ”āļ‚āļķāļ™āļˆāļēāļāļ›āļĢāļ°āļŠāļšāļāļēāļĢāļ“āđŒāļ•āļ­āļŠāļīāđˆāļ‡āļ—āļąāļ‡
                                            āđ‰                āđˆ āđ‰
 āļ›āļ§āļ‡ āļ„āļ·āļ­ āļ—āļąāđ‰āļ‡āļ˜āļĢāļĢāļĄāļŠāļēāļ•āļīāđāļĨāļ°āļˆāļīāļ•āđƒāļˆāļ­āļĒāđˆāļēāļ‡āđ€āļ›āđ‡āļ™āļŦāļ™āđˆāļ§āļĒāļĢāļ§āļĄāļ—āļĩāđˆāļĄāļĩāļ„āļ§āļēāļĄāļŦāļĄāļēāļĒ
 āļžāļĢāļ°āļžāļļāļ—āļ˜āļĻāļēāļŠāļ™āļēāļ•āļ­āļšāļ‚āđ‰āļ­āļāļģāļēāļŦāļ™āļ”āļ™āļĩāđ‰āđ„āļ”āđ‰.... āļ–āđ‰āļēāļˆāļ°āļĄāļĩāļĻāļēāļŠāļ™āļēāđƒāļ”āļ—āļĩāļĢāļąāļšāļĄāļ·āļ­āđ„āļ”āđ‰āļāļąāļš
                                                      āđˆ
 āļ„āļ§āļēāļĄāļ•āđ‰āļ­āļ‡āļāļēāļĢāļ—āļēāļ‡āļ§āļīāļ—āļĒāļēāļĻāļēāļŠāļ•āļĢāđŒāļŠāļĄāļąāļĒāđƒāļŦāļĄāđˆāļ›āļąāļˆāļˆāļļāļšāļ™ āļĻāļēāļŠāļ™āļēāļ™āļąāđ‰āļ™āļāđ‡āļ„āļ§āļĢāđ€āļ›āđ‡āļ™
                                          āļą
 āļžāļĢāļ°āļžāļļāļ—āļ˜āļĻāļēāļŠāļ™āļē"
  āļ­āđ‰āļēāļ‡āļ­āļīāļ‡ : http://www.watnai.org/einstein/The%20Human%20Side%20_st.html

                                                                           5
āļŦāļąāļ§āđƒāļˆāļ™āļąāļāļ›āļĢāļēāļŠāļāđŒ
                       āļŠāļļ āļˆāļī āļ›āļļ āļĨāļī

  āļŠāļļ : āļŠāļļāļ•āļ° (āļŸāļąāļ‡)
  āļˆāļī : āļˆāļīāļ™āļ•āļ° (āļ„āļīāļ”)
  āļ›āļļ : āļ›āļļāļˆāļ‰āļē (āļ–āļēāļĄ)
  āļĨāļī : āļĨāļīāļ‚āļīāļ• (āđ€āļ‚āļĩāļĒāļ™)



                                     6
Large Hadron Collider (LHC)




                              7
āļ„āļļāļ“āļĨāļąāļāļĐāļ“āļ°āļ‚āļ­āļ‡ LHC


 āļĨāļķāļāļĨāļ‡āđ„āļ›āđƒāļ•āđ‰āļ”āļīāļ™ 175 āđ€āļĄāļ•āļĢ
 āļĄāļĩāļĨāļąāļāļĐāļ“āļ°āđ€āļ›āđ‡āļ™āļ—āđˆāļ­āļ§āļ‡āļāļĨāļĄāļĢāļ°āļĒāļ°āļ—āļēāļ‡āļĒāļēāļ§ 27 āļāļīāđ‚āļĨāđ€āļĄāļ•āļĢ
 superconducting electromagnets āļ™āļąāđ‰āļ™āļ–āļđāļāļ—āļģāļēāđƒāļŦāđ‰āļĄāļĩāļ„āļ§āļēāļĄāđ€āļĒāđ‡āļ™āļ­āļĒāļđāđˆāļ—āļĩāđˆ -271 āļ­āļ‡āļĻāļē
 āđ€āļ‹āļĨāđ€āļ‹āļĩāļĒāļŠ āđ€āļžāļ·āļ­āļ—āļģāļēāđƒāļŦāđ‰āļĄāļąāļ™āđ€āļ›āđ‡āļ™āļ•āļąāļ§āļ™āļģāļēāļĒāļīāđˆāļ‡āļĒāļ§āļ” āļāđˆāļ­āļ™āļ—āļĩāđˆāļˆāļ°āđ€āļĢāļīāđˆāļĄāļ—āļģāļēāļāļēāļĢāļ—āļ”āļŠāļ­āļš
             āđˆ
 LHC āđƒāļŠāđ‰āļĢāļ°āļšāļš “Grid Computing” āđƒāļ™āļāļēāļĢāļ„āļģāļēāļ™āļ§āļ“āļ‚āđ‰āļ­āļĄāļđāļĨāļ—āļĩāļĄāļĩāļ›āļĢāļīāļĄāļēāļ“āļĄāļēāļāļ–āļķāļ‡
                                                 āđˆ
 15 Peta Byte (PB) āļ•āđˆāļ­āļ›āļĩ

                                                                           8
Avatar Movie




               9
Challenge of Avatar Movie
 One frame of certain shots, it's 24 frames per second,
  took 100 computer hours to render. Just one second
  was 2,400 hours.
 Weta operates a 10,000-square-foot facility that uses
  HP BL2x220c blades to process the effects for
  AVATAR and other films. The computing core
  contains some 40,000 processors and 104 terabytes
  of RAM.




                                                          10
11
References
 Shrek the Third
      http://www.linuxjournal.com/print/9653
 Kung Fu Panda digs grids
      http://www.isgtw.org/?pid=1000931
 Avatar
      Google Keyword "avatar movie render"




                                               12
Introduction to Computer Cluster




                                   13
High Performance Computing
 Supercomputers
 Computer Cluster
 Grid Computing
 Cloud Computing




                             14
Difference type of Cluster
 High Availability (HA)
 Load Balancing
 High Throughput Computing
 High Performance Computing




                              15
High-availability clustering
 Make more reliable computer system by having many
 computers working together and takeover when any of
 them fail




                                                       16
Computer Cluster Concept
 Cluster computing is a technology related to the
 building of high performance scalable computing
 system from a collection of small computing system
 and high speed interconnection network




                                                      17
High Throughput Computing



Jobs Queue




                             18
High Performance Computing



Sub task 1   Sub task 2
        Big job
Sub task 3   Sub task 4




                             19
Why Cluster ?
 One man show not work but Team Work!
   Research
      More complex and more amount data.
      Simulation.
 Technical
   Can't scale
      heat problem
      hard to design
      price


                                           20
Benefit of Cluster
 Low startup cost
 Scalability
 Rapid response, technology tracking
 More user-driven configuration
 Order of magnitude price-performance advantage




                                                  21
What Cluster do ?
 Sciences & Engineering
   Aerodynamics
   Bioinformatics
   Chemistry
   Oil and gas
   Weather prediction
   CFD
 Business
   Financial analysis
 Animation
   Graphic Rendering

                          22
HPC World

 Cluster 410 (83.4%)
 Constellation 2   ( 0.4%)
 MPP 88 (16.20%)




                             Data from http://www.top500.org
                                                 on 11/2009
                                                           23
Top500 : Architecture share 1993-2009




                                    24
Top500 : 11/2009 Application Share




                                     25
Top 10 on 11/2009




                    26
Today and Tomorrow




                     27
28
48 Core chip




               29
Gain Performance with Multi-core
 Multi-thread Programming
      Pthread
      OpenMP
 Message Passing
      PVM
      MPI
 Virtualization




                                   30
OpenMP with Visual C++




                         31
Example : OpenMP code.




     compile with “gcc -fopenmp source.c -o output”

                                                      32
Another Introduce,
 BI : Data Mining
 Machine Vision




                     33
Cloud Computing




                  34
Hot Issue

 Green Computing
 Virtualization
 On-Demand Computing
 Multi-threads / Parallel Programming




                                        35
What's Cloud Computing
 Cloud computing is Internet-based ("cloud")
   development and use of computer technology
   ("computing")




                                                36
37
Everything as a Service

 PaaS : Platform as a Service
 SaaS : Software as a Service
 IaaS : Infrastructure as a Service




                                      38
Players




          39
40
Amazon EC2 Pricing




                     41
Amazon EC2 Pricing (cont.)




                             42
Cloud Hosting




                43
Variety of Service




                     44
45
Weakness

 Reliability and security
      Outage (āđ„āļŸāļ”āļąāļš)
      Cost (āļ•āđ‰āļ™āļ—āļļāļ™)




                            46
Public / Private Cloud




                         47
Solution

 Usage public cloud with lower-priority applications
  such as absence reporting and expense reporting.
 Usage private cloud for privacy and security
 where companies will do some of their computing
  internally, possibly in a private cloud, while other
  tasks will be offloaded to the public cloud




                                                         48
āļ­āļĒāđˆāļēāļĒāļļāđˆāļ‡āļˆāļ™āļ‚āļ™āļēāļ”āļ™āļĩāđ‰




                    49
āļšāļēāļ‡āļ„āļĢāļąāđ‰āļ‡āļĨāļ­āļ‡āđ€āļ›āļĨāļĩāđˆāļĒāļ™āļ‡āļēāļ™āļ”āļđāļšāļēāļ‡āļ­āļēāļˆāļˆāļ°āļ”āļĩ
                        āđ‰




                                    50
Software Lists.
 Rocks Cluster (http://www.rocksclusters.org/)
 VirtualBox (http://www.virtualbox.org/)
 Ubuntu (http://www.ubuntu.com/)
 OpenMP (http://www.openmp.org/)
 Cloud Middleware
      Eucalyptus (http://www.eucalyptus.com/)
      OpenNebula (http://www.opennebula.org/)




                                                 51
About ClusterKit




                   52
Cluster Kit Co.,Ltd.

  Cluster Kit setting up on January 2007 (āļž.āļĻ. 2550),
    that have professional and experience team for
    design, implement and Training in Computer
    Cluster System and Grid Computing.




                                                        53
Strong Technical Skills

  Design
       High Performance System
       Load Balancing System
       High Availability System
  Application Software
       Web Load balancing
       MySQL Cluster
       Lustre File System
       {Linux, Windows} Cluster
       Rendering Farm

                                  54
Our proudly success
 Thai National Grid        Ramkhamhaeng
   Center (TNGC)             University
      “TERA” Cluster          “Sila” Cluster
        800 cores             286 cores
      Windows Cluster 33   BIOTEC
       nodes
                              “Eclipse” Cluster
                              704 cores




                                                  55
Our proudly success (contd.)

 Geo-Informatics and Space Technology Development
  Agency (GISTDA)
      Digital Thailand Project
      Web Load Balancing + Lustre


 Virgin Radio Thailand
      7 Nodes Web Load Balancing




                                                    56
ClusterKit Timeline




                      57
āļž.āļĻ.2547 (2004)
                  Asia Pacific Science &
                    Technology Facility
                    (APSTF)
                  A Research collaboration
                    between Kasetsart
                    University & Sun
                    Microsystems




                                             58
āļž.āļĻ.2548 (2005)

                  PAKSA Cluster for
                  medical research
                   – Avian Flu
                   – HIV
                  44 Sun Fire V20Z nodes
                  (88 Cores)




                                           59
āļž.āļĻ.2548 (2005)
 Geo-Informatics and Space Technology Development Agency (Public Organization)

 GISTDA
 6 nodes Pilot Project




                                                                                 60
āļž.āļĻ.2550 (2007)
 Cluster Kit Startup
 Installed the TERA Cluster @ ThaiGrid




                                         61
āļž.āļĻ. 2551 (2008)
 Full service at RU. (Sila Cluster)




                                      62
āļž.āļĻ.2552 (2009)
 Setup “Eclipse” Cluster at BIOTEC




                                     63
The End
 Q&A




          64

Weitere ÃĪhnliche Inhalte

Ähnlich wie HPC Cluster & Cloud Computing

āđ€āļ—āļ„āđ‚āļ™āđ‚āļĨāļĒāļĩ Cloud Computing
āđ€āļ—āļ„āđ‚āļ™āđ‚āļĨāļĒāļĩ Cloud Computingāđ€āļ—āļ„āđ‚āļ™āđ‚āļĨāļĒāļĩ Cloud Computing
āđ€āļ—āļ„āđ‚āļ™āđ‚āļĨāļĒāļĩ Cloud ComputingIMC Institute
 
Cloud Computing āļāļąāļšāļāļēāļĢāđƒāļŠāđ‰āļ‡āļēāļ™āđƒāļ™āļ­āļ‡āļ„āđŒāļāļĢāļ•āđˆāļēāļ‡āđ†
Cloud Computing āļāļąāļšāļāļēāļĢāđƒāļŠāđ‰āļ‡āļēāļ™āđƒāļ™āļ­āļ‡āļ„āđŒāļāļĢāļ•āđˆāļēāļ‡āđ†Cloud Computing āļāļąāļšāļāļēāļĢāđƒāļŠāđ‰āļ‡āļēāļ™āđƒāļ™āļ­āļ‡āļ„āđŒāļāļĢāļ•āđˆāļēāļ‡āđ†
Cloud Computing āļāļąāļšāļāļēāļĢāđƒāļŠāđ‰āļ‡āļēāļ™āđƒāļ™āļ­āļ‡āļ„āđŒāļāļĢāļ•āđˆāļēāļ‡āđ†Software Park Thailand
 
Cloud Computing āļāļąāļšāļāļēāļĢāđƒāļŠāđ‰āļ‡āļēāļ™āđƒāļ™āļ­āļ‡āļ„āđŒāļāļĢāļ•āđˆāļēāļ‡āđ†
Cloud Computing āļāļąāļšāļāļēāļĢāđƒāļŠāđ‰āļ‡āļēāļ™āđƒāļ™āļ­āļ‡āļ„āđŒāļāļĢāļ•āđˆāļēāļ‡āđ†Cloud Computing āļāļąāļšāļāļēāļĢāđƒāļŠāđ‰āļ‡āļēāļ™āđƒāļ™āļ­āļ‡āļ„āđŒāļāļĢāļ•āđˆāļēāļ‡āđ†
Cloud Computing āļāļąāļšāļāļēāļĢāđƒāļŠāđ‰āļ‡āļēāļ™āđƒāļ™āļ­āļ‡āļ„āđŒāļāļĢāļ•āđˆāļēāļ‡āđ†Thanachart Numnonda
 
Image colorization with artificial neural networks
Image colorization with artificial neural networksImage colorization with artificial neural networks
Image colorization with artificial neural networksTelexine
 
Python Programming for Lecturer_RUS_Nonthaburi 17may2019
Python Programming for Lecturer_RUS_Nonthaburi 17may2019Python Programming for Lecturer_RUS_Nonthaburi 17may2019
Python Programming for Lecturer_RUS_Nonthaburi 17may2019Asst.Prof.Dr.Arnut Ruttanatirakul
 
30879960 geospatial-system-on-cloud-computing
30879960 geospatial-system-on-cloud-computing30879960 geospatial-system-on-cloud-computing
30879960 geospatial-system-on-cloud-computingAyutthaya GIS
 
06 āļžāļ·āđ‰āļ™āļāļēāļ™āļĢāļ°āļšāļšāđ€āļ„āļĢāļ·āļ­āļ‚āđˆāļēāļĒāļ„āļ­āļĄāļžāļīāļ§āđ€āļ•āļ­āļĢāđŒ
06 āļžāļ·āđ‰āļ™āļāļēāļ™āļĢāļ°āļšāļšāđ€āļ„āļĢāļ·āļ­āļ‚āđˆāļēāļĒāļ„āļ­āļĄāļžāļīāļ§āđ€āļ•āļ­āļĢāđŒ06 āļžāļ·āđ‰āļ™āļāļēāļ™āļĢāļ°āļšāļšāđ€āļ„āļĢāļ·āļ­āļ‚āđˆāļēāļĒāļ„āļ­āļĄāļžāļīāļ§āđ€āļ•āļ­āļĢāđŒ
06 āļžāļ·āđ‰āļ™āļāļēāļ™āļĢāļ°āļšāļšāđ€āļ„āļĢāļ·āļ­āļ‚āđˆāļēāļĒāļ„āļ­āļĄāļžāļīāļ§āđ€āļ•āļ­āļĢāđŒteaw-sirinapa
 
Computer systemarchitecture
Computer systemarchitectureComputer systemarchitecture
Computer systemarchitectureSaranyu Srisrontong
 
Object oriented computing āļžāļĒāļļāļŦāļĒāļēāļ•āļĢāļēāļ‚āļ­āļ‡āđ€āļŦāļĨāđˆāļē objects
Object oriented computing āļžāļĒāļļāļŦāļĒāļēāļ•āļĢāļēāļ‚āļ­āļ‡āđ€āļŦāļĨāđˆāļē objectsObject oriented computing āļžāļĒāļļāļŦāļĒāļēāļ•āļĢāļēāļ‚āļ­āļ‡āđ€āļŦāļĨāđˆāļē objects
Object oriented computing āļžāļĒāļļāļŦāļĒāļēāļ•āļĢāļēāļ‚āļ­āļ‡āđ€āļŦāļĨāđˆāļē objectsSurapol Imi
 
51-307 Unit 1
51-307 Unit 151-307 Unit 1
51-307 Unit 1Naret Su
 

Ähnlich wie HPC Cluster & Cloud Computing (20)

āđ€āļ—āļ„āđ‚āļ™āđ‚āļĨāļĒāļĩ Cloud Computing
āđ€āļ—āļ„āđ‚āļ™āđ‚āļĨāļĒāļĩ Cloud Computingāđ€āļ—āļ„āđ‚āļ™āđ‚āļĨāļĒāļĩ Cloud Computing
āđ€āļ—āļ„āđ‚āļ™āđ‚āļĨāļĒāļĩ Cloud Computing
 
Cloud Computing āļāļąāļšāļāļēāļĢāđƒāļŠāđ‰āļ‡āļēāļ™āđƒāļ™āļ­āļ‡āļ„āđŒāļāļĢāļ•āđˆāļēāļ‡āđ†
Cloud Computing āļāļąāļšāļāļēāļĢāđƒāļŠāđ‰āļ‡āļēāļ™āđƒāļ™āļ­āļ‡āļ„āđŒāļāļĢāļ•āđˆāļēāļ‡āđ†Cloud Computing āļāļąāļšāļāļēāļĢāđƒāļŠāđ‰āļ‡āļēāļ™āđƒāļ™āļ­āļ‡āļ„āđŒāļāļĢāļ•āđˆāļēāļ‡āđ†
Cloud Computing āļāļąāļšāļāļēāļĢāđƒāļŠāđ‰āļ‡āļēāļ™āđƒāļ™āļ­āļ‡āļ„āđŒāļāļĢāļ•āđˆāļēāļ‡āđ†
 
Cloud Computing āļāļąāļšāļāļēāļĢāđƒāļŠāđ‰āļ‡āļēāļ™āđƒāļ™āļ­āļ‡āļ„āđŒāļāļĢāļ•āđˆāļēāļ‡āđ†
Cloud Computing āļāļąāļšāļāļēāļĢāđƒāļŠāđ‰āļ‡āļēāļ™āđƒāļ™āļ­āļ‡āļ„āđŒāļāļĢāļ•āđˆāļēāļ‡āđ†Cloud Computing āļāļąāļšāļāļēāļĢāđƒāļŠāđ‰āļ‡āļēāļ™āđƒāļ™āļ­āļ‡āļ„āđŒāļāļĢāļ•āđˆāļēāļ‡āđ†
Cloud Computing āļāļąāļšāļāļēāļĢāđƒāļŠāđ‰āļ‡āļēāļ™āđƒāļ™āļ­āļ‡āļ„āđŒāļāļĢāļ•āđˆāļēāļ‡āđ†
 
Image colorization with artificial neural networks
Image colorization with artificial neural networksImage colorization with artificial neural networks
Image colorization with artificial neural networks
 
Docker and Software Framework_24 may16_rus_DR.ARNUT
Docker and Software Framework_24 may16_rus_DR.ARNUTDocker and Software Framework_24 may16_rus_DR.ARNUT
Docker and Software Framework_24 may16_rus_DR.ARNUT
 
Python Programming for Lecturer_RUS_Nonthaburi 17may2019
Python Programming for Lecturer_RUS_Nonthaburi 17may2019Python Programming for Lecturer_RUS_Nonthaburi 17may2019
Python Programming for Lecturer_RUS_Nonthaburi 17may2019
 
Emerging technologies in teaching and learning
Emerging technologies in teaching and learningEmerging technologies in teaching and learning
Emerging technologies in teaching and learning
 
Emerging tech in teaching and learning
Emerging tech in teaching and learningEmerging tech in teaching and learning
Emerging tech in teaching and learning
 
30879960 geospatial-system-on-cloud-computing
30879960 geospatial-system-on-cloud-computing30879960 geospatial-system-on-cloud-computing
30879960 geospatial-system-on-cloud-computing
 
e-Government Cloud Computing
e-Government Cloud Computinge-Government Cloud Computing
e-Government Cloud Computing
 
Cloud computing
Cloud computingCloud computing
Cloud computing
 
IT Skills for Teacher path 1
IT Skills for Teacher path 1IT Skills for Teacher path 1
IT Skills for Teacher path 1
 
06 āļžāļ·āđ‰āļ™āļāļēāļ™āļĢāļ°āļšāļšāđ€āļ„āļĢāļ·āļ­āļ‚āđˆāļēāļĒāļ„āļ­āļĄāļžāļīāļ§āđ€āļ•āļ­āļĢāđŒ
06 āļžāļ·āđ‰āļ™āļāļēāļ™āļĢāļ°āļšāļšāđ€āļ„āļĢāļ·āļ­āļ‚āđˆāļēāļĒāļ„āļ­āļĄāļžāļīāļ§āđ€āļ•āļ­āļĢāđŒ06 āļžāļ·āđ‰āļ™āļāļēāļ™āļĢāļ°āļšāļšāđ€āļ„āļĢāļ·āļ­āļ‚āđˆāļēāļĒāļ„āļ­āļĄāļžāļīāļ§āđ€āļ•āļ­āļĢāđŒ
06 āļžāļ·āđ‰āļ™āļāļēāļ™āļĢāļ°āļšāļšāđ€āļ„āļĢāļ·āļ­āļ‚āđˆāļēāļĒāļ„āļ­āļĄāļžāļīāļ§āđ€āļ•āļ­āļĢāđŒ
 
Computer systemarchitecture
Computer systemarchitectureComputer systemarchitecture
Computer systemarchitecture
 
IT Skills for Teacher path 2
IT Skills for Teacher path 2IT Skills for Teacher path 2
IT Skills for Teacher path 2
 
Cloud computing
Cloud computingCloud computing
Cloud computing
 
Proposal
ProposalProposal
Proposal
 
Object oriented computing āļžāļĒāļļāļŦāļĒāļēāļ•āļĢāļēāļ‚āļ­āļ‡āđ€āļŦāļĨāđˆāļē objects
Object oriented computing āļžāļĒāļļāļŦāļĒāļēāļ•āļĢāļēāļ‚āļ­āļ‡āđ€āļŦāļĨāđˆāļē objectsObject oriented computing āļžāļĒāļļāļŦāļĒāļēāļ•āļĢāļēāļ‚āļ­āļ‡āđ€āļŦāļĨāđˆāļē objects
Object oriented computing āļžāļĒāļļāļŦāļĒāļēāļ•āļĢāļēāļ‚āļ­āļ‡āđ€āļŦāļĨāđˆāļē objects
 
Deep Learning Smart Bin
Deep Learning Smart BinDeep Learning Smart Bin
Deep Learning Smart Bin
 
51-307 Unit 1
51-307 Unit 151-307 Unit 1
51-307 Unit 1
 

HPC Cluster & Cloud Computing

  • 1. HPC Cluster & Cloud Computing āļāļīāļ•āļ•āļīāļĢāļąāļāļĐāđŒ āļĄāđˆāļ§āļ‡āļĄāļīāđˆāļ‡āļŠāļļāļ‚ (Kittirak Moungmingsuk) kittirak@clusterkit.co.th Feb 9, 2010 @ NIDA
  • 2. What's ? E=mc2 2
  • 3. Mass–energy equivalence E : Energy (āļžāļĨāļąāļ‡āļ‡āļēāļ™) m : mass (āļĄāļ§āļĨ) c : āļ„āļ§āļēāļĄāđ€āļĢāđ‡āļ§āđāļŠāļ‡ (āļĄāļēāļˆāļēāļāļ āļēāļĐāļēāļĨāļ°āļ•āļīāļ™āļ„āļģāļēāļ§āđˆāļē celeritas āđāļ›āļĨāļ§āđˆāļē āļ­āļąāļ•āļĢāļēāđ€āļĢāđ‡āļ§) 3
  • 5. āđāļ›āļĨāļˆāļēāļāļ•āļ­āļ™āļŦāļ™āļķāđˆāļ‡āđƒāļ™ “The Human Side” "āļĻāļēāļŠāļ™āļē āđƒāļ™āļ­āļ™āļēāļ„āļ• āļˆāļ°āļ•āđ‰āļ­āļ‡āđ€āļ›āđ‡āļ™āļĻāļēāļŠāļ™āļēāļŠāļēāļāļĨ āļĻāļēāļŠāļ™āļēāļ™āļąāđ‰āļ™āļ„āļ§āļĢāļ­āļĒāļđāđˆāđ€āļŦāļ™āļ·āļ­ āļžāļĢāļ°āđ€āļˆāđ‰āļēāļ—āļĩāļĄāļĩāļ•āļąāļ§āļ•āļ™ āđāļĨāļ°āļ„āļ§āļĢāļˆāļ°āđ€āļ§āđ‰āļ™āļ„āļģāļēāļŠāļ­āļ™āđāļšāļšāļŠāļīāļ—āļ˜āļąāļ™āļ•āđŒ (āļ„āļ·āļ­āđ€āļ›āđ‡āļ™āđāļšāļš āđˆ āļŠāļģāļēāđ€āļĢāđ‡āļˆāļĢāļđāļ›āļ—āļĩāđˆāđƒāļŦāđ‰āđ€āļŠāļ·āđˆāļ­āļ•āļēāļĄāđ€āļžāļĩāļĒāļ‡āļ­āļĒāđˆāļēāļ‡āđ€āļ”āļĩāļĒāļ§) āđāļĨāļ°āđāļšāļšāđ€āļ—āļ§āļ§āļīāļ—āļĒāļē(āļ„āļ·āļ­āļžāļķāļ‡āđ€āļ—āļ§āļ”āļē āđˆ āđ€āļ›āđ‡āļ™āļŦāļĨāļąāļāđƒāļŦāļāđˆ) āļĻāļēāļŠāļ™āļēāļ™āļąāđ‰āļ™ āđ€āļĄāļ·āđˆāļ­āļ„āļĢāļ­āļšāļ„āļĨāļļāļĄāļ—āļąāđ‰āļ‡āļ˜āļĢāļĢāļĄāļŠāļēāļ•āļīāđāļĨāļ°āļˆāļīāļ•āđƒāļˆ āļˆāļķāļ‡āļ„āļ§āļĢāļĄāļĩ āļĢāļēāļāļāļēāļ™āļ­āļĒāļđāđˆāļšāļ™āļŠāļēāļĄāļąāļāļŠāļģāļēāļ™āļķāļāļ—āļēāļ‡āļĻāļēāļŠāļ™āļē āļ—āļĩāđˆāđ€āļāļīāļ”āļ‚āļķāļ™āļˆāļēāļāļ›āļĢāļ°āļŠāļšāļāļēāļĢāļ“āđŒāļ•āļ­āļŠāļīāđˆāļ‡āļ—āļąāļ‡ āđ‰ āđˆ āđ‰ āļ›āļ§āļ‡ āļ„āļ·āļ­ āļ—āļąāđ‰āļ‡āļ˜āļĢāļĢāļĄāļŠāļēāļ•āļīāđāļĨāļ°āļˆāļīāļ•āđƒāļˆāļ­āļĒāđˆāļēāļ‡āđ€āļ›āđ‡āļ™āļŦāļ™āđˆāļ§āļĒāļĢāļ§āļĄāļ—āļĩāđˆāļĄāļĩāļ„āļ§āļēāļĄāļŦāļĄāļēāļĒ āļžāļĢāļ°āļžāļļāļ—āļ˜āļĻāļēāļŠāļ™āļēāļ•āļ­āļšāļ‚āđ‰āļ­āļāļģāļēāļŦāļ™āļ”āļ™āļĩāđ‰āđ„āļ”āđ‰.... āļ–āđ‰āļēāļˆāļ°āļĄāļĩāļĻāļēāļŠāļ™āļēāđƒāļ”āļ—āļĩāļĢāļąāļšāļĄāļ·āļ­āđ„āļ”āđ‰āļāļąāļš āđˆ āļ„āļ§āļēāļĄāļ•āđ‰āļ­āļ‡āļāļēāļĢāļ—āļēāļ‡āļ§āļīāļ—āļĒāļēāļĻāļēāļŠāļ•āļĢāđŒāļŠāļĄāļąāļĒāđƒāļŦāļĄāđˆāļ›āļąāļˆāļˆāļļāļšāļ™ āļĻāļēāļŠāļ™āļēāļ™āļąāđ‰āļ™āļāđ‡āļ„āļ§āļĢāđ€āļ›āđ‡āļ™ āļą āļžāļĢāļ°āļžāļļāļ—āļ˜āļĻāļēāļŠāļ™āļē" āļ­āđ‰āļēāļ‡āļ­āļīāļ‡ : http://www.watnai.org/einstein/The%20Human%20Side%20_st.html 5
  • 6. āļŦāļąāļ§āđƒāļˆāļ™āļąāļāļ›āļĢāļēāļŠāļāđŒ āļŠāļļ āļˆāļī āļ›āļļ āļĨāļī āļŠāļļ : āļŠāļļāļ•āļ° (āļŸāļąāļ‡) āļˆāļī : āļˆāļīāļ™āļ•āļ° (āļ„āļīāļ”) āļ›āļļ : āļ›āļļāļˆāļ‰āļē (āļ–āļēāļĄ) āļĨāļī : āļĨāļīāļ‚āļīāļ• (āđ€āļ‚āļĩāļĒāļ™) 6
  • 8. āļ„āļļāļ“āļĨāļąāļāļĐāļ“āļ°āļ‚āļ­āļ‡ LHC āļĨāļķāļāļĨāļ‡āđ„āļ›āđƒāļ•āđ‰āļ”āļīāļ™ 175 āđ€āļĄāļ•āļĢ āļĄāļĩāļĨāļąāļāļĐāļ“āļ°āđ€āļ›āđ‡āļ™āļ—āđˆāļ­āļ§āļ‡āļāļĨāļĄāļĢāļ°āļĒāļ°āļ—āļēāļ‡āļĒāļēāļ§ 27 āļāļīāđ‚āļĨāđ€āļĄāļ•āļĢ superconducting electromagnets āļ™āļąāđ‰āļ™āļ–āļđāļāļ—āļģāļēāđƒāļŦāđ‰āļĄāļĩāļ„āļ§āļēāļĄāđ€āļĒāđ‡āļ™āļ­āļĒāļđāđˆāļ—āļĩāđˆ -271 āļ­āļ‡āļĻāļē āđ€āļ‹āļĨāđ€āļ‹āļĩāļĒāļŠ āđ€āļžāļ·āļ­āļ—āļģāļēāđƒāļŦāđ‰āļĄāļąāļ™āđ€āļ›āđ‡āļ™āļ•āļąāļ§āļ™āļģāļēāļĒāļīāđˆāļ‡āļĒāļ§āļ” āļāđˆāļ­āļ™āļ—āļĩāđˆāļˆāļ°āđ€āļĢāļīāđˆāļĄāļ—āļģāļēāļāļēāļĢāļ—āļ”āļŠāļ­āļš āđˆ LHC āđƒāļŠāđ‰āļĢāļ°āļšāļš “Grid Computing” āđƒāļ™āļāļēāļĢāļ„āļģāļēāļ™āļ§āļ“āļ‚āđ‰āļ­āļĄāļđāļĨāļ—āļĩāļĄāļĩāļ›āļĢāļīāļĄāļēāļ“āļĄāļēāļāļ–āļķāļ‡ āđˆ 15 Peta Byte (PB) āļ•āđˆāļ­āļ›āļĩ 8
  • 10. Challenge of Avatar Movie One frame of certain shots, it's 24 frames per second, took 100 computer hours to render. Just one second was 2,400 hours. Weta operates a 10,000-square-foot facility that uses HP BL2x220c blades to process the effects for AVATAR and other films. The computing core contains some 40,000 processors and 104 terabytes of RAM. 10
  • 11. 11
  • 12. References Shrek the Third http://www.linuxjournal.com/print/9653 Kung Fu Panda digs grids http://www.isgtw.org/?pid=1000931 Avatar Google Keyword "avatar movie render" 12
  • 14. High Performance Computing Supercomputers Computer Cluster Grid Computing Cloud Computing 14
  • 15. Difference type of Cluster High Availability (HA) Load Balancing High Throughput Computing High Performance Computing 15
  • 16. High-availability clustering Make more reliable computer system by having many computers working together and takeover when any of them fail 16
  • 17. Computer Cluster Concept Cluster computing is a technology related to the building of high performance scalable computing system from a collection of small computing system and high speed interconnection network 17
  • 19. High Performance Computing Sub task 1 Sub task 2 Big job Sub task 3 Sub task 4 19
  • 20. Why Cluster ? One man show not work but Team Work! Research More complex and more amount data. Simulation. Technical Can't scale heat problem hard to design price 20
  • 21. Benefit of Cluster Low startup cost Scalability Rapid response, technology tracking More user-driven configuration Order of magnitude price-performance advantage 21
  • 22. What Cluster do ? Sciences & Engineering Aerodynamics Bioinformatics Chemistry Oil and gas Weather prediction CFD Business Financial analysis Animation Graphic Rendering 22
  • 23. HPC World Cluster 410 (83.4%) Constellation 2 ( 0.4%) MPP 88 (16.20%) Data from http://www.top500.org on 11/2009 23
  • 24. Top500 : Architecture share 1993-2009 24
  • 25. Top500 : 11/2009 Application Share 25
  • 26. Top 10 on 11/2009 26
  • 28. 28
  • 30. Gain Performance with Multi-core Multi-thread Programming Pthread OpenMP Message Passing PVM MPI Virtualization 30
  • 32. Example : OpenMP code. compile with “gcc -fopenmp source.c -o output” 32
  • 33. Another Introduce, BI : Data Mining Machine Vision 33
  • 35. Hot Issue Green Computing Virtualization On-Demand Computing Multi-threads / Parallel Programming 35
  • 36. What's Cloud Computing Cloud computing is Internet-based ("cloud") development and use of computer technology ("computing") 36
  • 37. 37
  • 38. Everything as a Service PaaS : Platform as a Service SaaS : Software as a Service IaaS : Infrastructure as a Service 38
  • 39. Players 39
  • 40. 40
  • 42. Amazon EC2 Pricing (cont.) 42
  • 45. 45
  • 46. Weakness Reliability and security Outage (āđ„āļŸāļ”āļąāļš) Cost (āļ•āđ‰āļ™āļ—āļļāļ™) 46
  • 47. Public / Private Cloud 47
  • 48. Solution Usage public cloud with lower-priority applications such as absence reporting and expense reporting. Usage private cloud for privacy and security where companies will do some of their computing internally, possibly in a private cloud, while other tasks will be offloaded to the public cloud 48
  • 51. Software Lists. Rocks Cluster (http://www.rocksclusters.org/) VirtualBox (http://www.virtualbox.org/) Ubuntu (http://www.ubuntu.com/) OpenMP (http://www.openmp.org/) Cloud Middleware Eucalyptus (http://www.eucalyptus.com/) OpenNebula (http://www.opennebula.org/) 51
  • 53. Cluster Kit Co.,Ltd. Cluster Kit setting up on January 2007 (āļž.āļĻ. 2550), that have professional and experience team for design, implement and Training in Computer Cluster System and Grid Computing. 53
  • 54. Strong Technical Skills Design High Performance System Load Balancing System High Availability System Application Software Web Load balancing MySQL Cluster Lustre File System {Linux, Windows} Cluster Rendering Farm 54
  • 55. Our proudly success Thai National Grid Ramkhamhaeng Center (TNGC) University “TERA” Cluster “Sila” Cluster 800 cores 286 cores Windows Cluster 33 BIOTEC nodes “Eclipse” Cluster 704 cores 55
  • 56. Our proudly success (contd.) Geo-Informatics and Space Technology Development Agency (GISTDA) Digital Thailand Project Web Load Balancing + Lustre Virgin Radio Thailand 7 Nodes Web Load Balancing 56
  • 58. āļž.āļĻ.2547 (2004) Asia Pacific Science & Technology Facility (APSTF) A Research collaboration between Kasetsart University & Sun Microsystems 58
  • 59. āļž.āļĻ.2548 (2005) PAKSA Cluster for medical research – Avian Flu – HIV 44 Sun Fire V20Z nodes (88 Cores) 59
  • 60. āļž.āļĻ.2548 (2005) Geo-Informatics and Space Technology Development Agency (Public Organization) GISTDA 6 nodes Pilot Project 60
  • 61. āļž.āļĻ.2550 (2007) Cluster Kit Startup Installed the TERA Cluster @ ThaiGrid 61
  • 62. āļž.āļĻ. 2551 (2008) Full service at RU. (Sila Cluster) 62
  • 63. āļž.āļĻ.2552 (2009) Setup “Eclipse” Cluster at BIOTEC 63