Grid

1
Grid Globus

• ???
• ???

2
Part 1 Grid
3
A Story Late 80s, early 90s

• Gigabit testbeds program
• CASA (southwest)
• MAGIC and BLANCA (Midwest)
• AURORA and NECTAR (northeast)
• VISTANET (southeast)
• Focus on connection and bandwidth

4
Supercomputing 95

• I-WAY, the first large-scale Grid experiment
• Consisted of a Grid of 17 sites connected by vBNS
  
• Over 60 applications ran on the I-WAY during
  SC95
• Users could use single authentication and job
  submission across multiple sites or they could
  work directly with end-users

5
Mid to late 90s

• Academic software projects (e.g., Legion, Globus)
• Application experiments (e.g., IPG, ASCI,
  National Technology Grid)

To be continued
6
What is Grid
7
Electrical power grid
? Grid

• Power plant (dynamo)

• HPCC ( high performance computer)

Knowledge, transaction
Electricity
Electrical network
WAN, LAN
Appliances (applications)
Terminal (applications)
8
Electrical power grid

• Dependable
• Electricity is always there. When the switch is
  thrown, the light always comes on.
• Consistent
• The uniform interface means that all appliances
  use the same connector.
• Pervasive
• Electrical service is virtually everywhere,
  throughout the world.

9
Grid

• Dependable
• Can Provide performance and functionality
  guarantees
• Consistent
• Uniform interfaces to a wide variety of
  resources
• Pervasive
• Ability to Plug In from anywhere

10
What is Grid ?

• Enabling the coordinated use of geographically
• distributed resources in the absence of
• central control, omniscience, strong trust
• relationships
• Ian
  Foster (Globus)
• What, exactly, is the Grid ?
• http//access.ncsa.uiuc.edu/CoverStories/WhatisGri
  d/

11
What is special ?

• High performance
• Bandwidth, speed, architecture
• Shared resources
• Computer, software, data, instrument
• Single system image
• single, coherent virtual machine
• Knowledge generating

12
Taxonomy
Distributed Supercomputing
Computational Grid
High Throughput
Grid
Data Grid
On Demand
Collaborative
Service Grid
Multimedia
13
Computational Grid

• The computational Grid category denotes
  systems that have a higher aggregate
  computational capacity available for single
  applications than the capacity of any constituent
  machine in the system.

Back
14
Data Grid

• The data Grid category is for systems that
  provide an infrastructure for synthesizing new
  information from data repositories such as
  digital libraries or data warehouses that are
  distributed in a wide area network.

TIDE The Tele-Immersive Data Explorer (USA)
Back
15
Service Grid

• The service Grid category is for systems that
  provide services that are not provided by any
  single machine.

Back
16
Distributed Supercomputing

• Distributed Particle
• Physics Research
• California Institute of Technology, USA
• CERN

Back
17
High Throughput
SETI_at_home the Search of ExtraTerrestrial
Intelligence
IBM ASCI White 12.3 TeraFLOPs/sec
Back
18
On Demand

• Advanced Networking
• for Telemicroscopy
• National Center for Microscopy and Imaging
  Research , UCSD, USA
• SDSC, USA
• Research Center for Ultra-High Voltage Electron
  Microscopy, Osaka Univ., Japan
• NLANR, USA
• Lawrence Berkeley National Laboratory, USA

Back
19
Collaborative

• CyberCAD Internet Distributed Interactive
  Collaborative Design
• Temasek Polytechnic, Singapore
• National University of Singapore
• Indiana University, USA

Back
20
Multimedia

• MediaZine?A Combination of Television, WWW,
• Telecommunications and 3D Computer Graphics
• Fraunhofer Institute Graphische Datenverarbeitung
  (IGD), Germany
• Centre for Advanced Media Technology (CAM Tech),
  Singapore

21
Grid vs Traditional distributed computing

• Heterogeneous hardware environment
• computing platforms
• network connections
• storage systems and caches
• Wide-area distribution
• Wide-area network latency and bandwidth
• Resources in different administration domains
• Dynamic environment
• Resources enter and leave grid

22
Research areas

• Architecture (Jini)
• Communication and remote access to data
• Information services
• Resource management
• Tasks scheduling
• Performance
• User management
• Security
• User Environment (programming model)
• Applications

23
Communication models

• Client-Server Model,
  Remote procedure call,
  Group communication
• In a grid
• Algorithms must tolerate wide-area latency for
  message transfers
• Avoid large numbers of messages
• Typically perform larger transfers, initiate
  remote jobs rather than procedure calls

24
Synchronization

• Clock synchronization, Election
  algorithms determine a coordinator,
  Atomic transactions
• In a grid
• With wide-area latencies, typically perform
  synchronization on larger grain
• Can implement atomic operations

25
Processes and Processors

• Threads,
  Allocating Processors,
  Scheduling and co-scheduling
  resources, Fault tolerance
• In a grid
  scheduling, allocation, fault tolerance
  issues get more complicated in the wide area
  environment

26
Distributed file systems

• File service (read, write, controls access),
  Creating,
  deleting managing directories, Naming, Sharing,
  Caching
  and consistency,
  Replication and updates
• In a grid
  same issues complicated by wide
  area distribution, different administrative
  domains, enormous data sets

27
Some projects

• Globus
• Legion
• IPG (Information Power Grid)
• TeraGrid
• European DataGrid
• NHPCE
  (National High Performance Computing
  Environment)
• Peer to Peer computing

28
Globus

• Ian Foster (ANL/UC) and
  Carl Kesselman (USC/ISI)
• Bag of services  model
• Globus Metacomputing Toolkit (GMT) GRAM, Nexus,
  MDS, GSI, HBM, GASS
• Globus Ubiquitous Supercomputing Testbed
  Organization (GUSTO).
• http//www.globus.org

29
Legion

• Andrew Grimshaw (UVA)
• Single, coherent virtual machine model
• Everything is an object (all hardware and
  software components)
• Interface and basic functionality a set of core
  object types
• Users can define their own class
• Centurion and the NPACI-net testbeds
• http//legion.virginia.edu/

30
Information Power Grid

• NASA
• Large-scale science and engineering
• Globus providing the Grid Common Services
• Computing resources 800 CPU nodes in half a
  dozen SGI Origin 2000s and several workstation
  clusters
• WAN interconnects of at least 100 mbit/s
• Storage resources 50-100 Terabytes of archival
  information/data storage
• http//www.ipg.nasa.gov/

31
TeraGrid (2001.8)

• NSF (NCSA, SDSC, ANL, CIT), IBM
• The world's largest, fastest, most comprehensive,
  distributed infrastructure
• Built from four individual clusters
• 53 million
• 13.6 teraflops
• 450 terabytes
• 40 gigabits/second
• http//www.teragrid.org/

32
European DataGrid

• CERN The European Organisation for Nuclear
  Research (20 European countries. 2,700 staff,
  6,000 users)
• Project objectives
• Middleware for fabric Grid management
• Large scale testbed
• Production quality demonstrations
• http//www.eu-datagrid.org

33
NHPCE

• National High Performance Computing Environment
• Beijing, Shanghai, Hefei, Xian, Changsha,
  Chengdu
• Infrastructure, Gridware
• Weather forecast, Bioinformaion, Industry
  applications, Information services
• http//www.grid.org.cn/

34
Peer-to-peer computing

• Intel (Marquam)
• The sharing of computer resources and services by
  direct exchange between systems.
• Information, processing cycles, cache storage,
  and disk storage for files
• Collaboration, Edge services, Distributed
  computing and resources, Intelligent agents.
• http//www.peer-to-peerwg.org/

35
Lists

• http//www.gridforum.org/info/Initiatives.html
• http//www.globus.org/about/related.html
• http//www.gridcomputing.com/

36
Part2 Globus
37
Globus Project

• 1996,Argonne National Lab.

-Basic research in grid-related
technologies -Development of Globus
Toolkit -Construction of production grids
testbeds -Application experiments

• Globus Toolkit 1.1.3---Jun 2000
• Globus Toolkit 1.1.4---Nov 2000

• TeraGrid uses globus as a core software

38
Globus Layered Architecture
Applications
High-level Services and Tools
GlobusView
Testbed Status
DUROC
globusrun
MPI
Nimrod/G
Condor
HPC
Core Services
GRAM
Nexus
Metacomputing Directory Service
Globus Security Interface
Heartbeat Monitor
I/O
GASS
Local Services
Condor
MPI
TCP
UDP
LSF
NQE
PBS
AIX
Linux
Solaris
39
Globus Toolkit Services

• Security
• Grid Security Infrastructure (GSI)
• Information Service
• Grid Information System (GIS)
• Resource Management
• Globus Resource Allocation Manager (GRAM)
• Data Management
• Globus Access to Secondary Storage (GASS)
• Communication
• NEXUS, Globus_io
• Process Monitoring
• Heart Beat Monitor (HBM)

40
1. Security

• GSI Grid Security Infrastructure
• Uses GSSAPI (a Generic Security Services API
  implementation Based on SSL)
• Single sign-on for all resources
• Mutual authentication

41
GSI User Management

• Grid user id,certification and key
• Example
• grid-cert-request
• grid-proxy-init
• Enter PEM pass phrase
• Gridmap file
• "/OGlobus/ONPACI/OUSDSC/CNRich Gallup rpg

42
GSI - Secure Remote Startup
1. Exchange certificates 2. Check gridmap
file 3. Lookup Service 4. Run service program

43
2. Information Service

• GIS Grid Information System
• Based on LDAPv3
• Critical for Grid
• - Resource Discovery
• What resources are available?
• - Resource Selection
• What is the state of the grid?
• - Application Configuration and Adaptation
• How to optimize resource use

44
GIS Structure Classic

• Centralized
• Use the same server mds.globus.org

45
GIS Structure Standard (1)

• Decentralized
• -GRIS and GIIS
• Resource Description Services --
• Grid Resource Information Service (GRIS)
• - One GRIS server running on each resource
• - Provide the state of the local resource

46
GIS Structure Standard (2)

• Aggregate Directory Services --
• Grid Index Information Service (GIIS)
• - Gathers information from multiple GRIS
  servers
• - Each GIIS is optimized for particular queries

47
3. Resource Management
48
Major Components (1)

• RSL Resource Specification Language
• - Information exchange for components
• - Resource requirements Job configuration
• Example
• (countgt5) (countlt10)
• (max_time240) (memorygt64)
• (executablemyprog)

49
Major Components (2)

• GRAM Globus Resource Allocation Manager
• -Processes requests for execution on remote
  resources
• -Allocates local resources required by remote
  requests
• -Provides an API for submit job,jobs status
  query and cancel job

50
Major Components (3)

• DUROC Dynamically Updated Request Online
  Co-allocator
• - Co-allocationSimultaneous allocation of a
  resource set
• Example
• ( (count5)(memorygt64)
• (executablep1))
• ((networkatm) (executablep2))

51
Major Components (4)

• GARA General-purpose Architecture for
  Reservation and Allocation
• - 2nd generation resource management service
• - Advance reservation of resources
• - A research prototype currently

52
Interesting Things

• Broker and Meta-Scheduler
• - No implementation in Globus Toolkit
• - Nimrod/G, AppLeS build based on GRAM and GIS
• - PBS,LSF
• Future
• -Reservation,Failure Management,Security,XML

53
4. Data Management

• GASS
• Globus Access to Secondary Storage
• -For GRAM to access remote data
• GridFTP
• -For high-performance, reliable data transfer in
  the Grid environment
• Replica Management
• -A map between logical to one or more physical

54
5. Communication (1)

• NEXUS
• 1. Uni-API MultiMethod Communication
• -- Where?What?When?
• -- Unicast vs. multicast
• Reliable vs. unreliable
• Quality of service
• Encrypted vs. unencrypted
• Compressed vs. uncompressed
• -- automatically or manually

55
Communication (2)

• 2. Communication Links (CLs)
• 3. Remote Service Request (RSR)
• 4. MPI-G implementation based on NEXUS

56
Communication (3)

• Globus_io
• -Compatibility?Security?QoS
• -- Easier Win32 portability
• -- Connection properties query
• -MPI-G implementation based on Globus_io
• Future
• NEXUS?Globus_io

57
An Example Installation
58
Useful Site

• www.globus.org
• discuss_at_globus.org

59
End

• Thank you !

60
?????

• ???????,???????????
• ??????,??????
• ???? BBS ?
  HPCC ?
• ????FTP//202.38.76.228/incoming/Grid