1 July 30, 2005 Grid Computing Principles Consortium for Computational Science and High Performance...

1

July 30, 2005

Grid Computing Principles

Consortium for Computational Science and High Performance Computing

2005 Summer Workshop, July 29-July 31, 2005

2

Grid Computing Coursework Development Team

UNC-Charlotte

Barry Wilkinson

Kevin Hammond(PhD Student)

Western CarolinaUniversity

Mark Holliday

James Ruff (Undergraduate student)

Elon University

Joel Hollingsworth

Appalachian State University: Darryl Cook Systems Administrator

3

Introduction to Grid Computing

8:30 am - 9:45 am

Barry WilkinsonDepartment of Computer Science

UNC-Charlotte

4

Need to harness computersOriginal driving force behind grid computing the same as behind the early development of networks that became the Internet:

– Connecting computers at distributed sites for high performance computing.

• However, just as the Internet has changed, grid computing has changed to embrace collaborative computing.

5

History

• Began in mid 1990’s with experiments using computers at geographically dispersed sites.

• Seminal experiment – “I-way” experiment at 1995 Supercomputing conference (SC’95), using 17 sites across the US running:– 60+ applications.– Existing networks (10 networks).

6

1995 2000 200519901985

Distributed computing

Remote Procedure calls (RPC)Concept of service registry

Beginnings of service oriented architecture

Object oriented approachesJava Remote Method Invocation (RMI)

CORBA (Common Request Broker Architecture)

Cluster computing

Software Techniques:

Computing platforms:

Parallel computers

Geographically distributed computers (Grid computing in the broadest sense)

Web services

SC’95 experiment

7

Grid Computing

• Using distributed computers and resources collectively.

• Usually associated with geographically distributed computers and resources on a special high speed network, or the Internet.

• Now become much more that last slide suggests.

8

Shared Resources

Can share much more than just computers:

• Storage• Sensors for experiments at particular sites• Application Software• Databases• Network capacity, …

9

Computational Grid Applications

• Biomedical research

• Industrial research

• Engineering research

• Studies in Physics and Chemistry

10

Sample Grid Computing Projects

Physical Sciences:• Large Hadron Collider project (CERN)• DOE Particle Physics Data grid• DOE Science grid• AstroGrid• Comb-e-Chem project

Natural and Life sciences:• Protein Data grid• Mcell project

Engineering Design:• Distributed Aircraft Maintenance Environment• NASA Information Power grid

11

Science Today

is a Team Sport

I. Foster

12

eScience

eScience [n]: Large-scale science carried out through distributed collaborations—often leveraging access to large-scale data & computing

I. Foster

NSF Network for Earthquake Engineering Simulation (NEES)

Transform our ability to carry out research vital to reducing vulnerability to catastrophic earthquakes

I. Foster

http://www.nees.org/EQ/sites/images/UCSD_72A.JPG

http://www.nees.org/EQ/sites/images/lehigh_72a.jpg

http://www.nees.org/EQ/sites/images/uiuc_72A.jpg

http://www.nees.org/EQ/sites/images/cornell_72a.jpg

http://www.nees.org/EQ/sites/images/BYU_72A.JPG

Global Knowledge Communities: e.g., High Energy Physics

I. Foster

15www.earthsystemgrid.org

DOE Earth System Grid

Goal: address technical obstacles to the sharing & analysis of high-volume data from advanced earth system models

I. Foster

16

Earth System Grid I. Foster

17

TeraGridFunded by NSF in 2002 to link 5 supercomputer sites

with 40 Gb/s links

18

TeraGrid

19

Grid networks for collaborative grid computing projects

Grids have been set up at the local level, national level and international level throughout the world, to promote grid computing

20

CiscoEPA

North Carolina’s Foundation for Grid: NCREN

4-7 MCNC-owned Clusters distributed throughout the stateLocations still under evaluation

Internet Internet 2

NLR

Internet Internet 2

NLR

InternetInternet

Existing: Blend of owned and leased fiber and circuits moving toward resilient rings powered by Cisco routers

Planned: Strong focus on owned and leased fiber, Lambda, and few circuits, in resilient rings powered by Cisco routers and Wave Division Multiplexers

Close to home:

From “Grid Computing in the Industry” by Wolfgang Gentzsch, presentation to Fall 2004 grid computing course. Full set of slides on course home page.

21

Grid2003: An Operational National Grid28 sites: Universities + national labs2800 CPUs, 400–1300 jobsRunning since October 2003Applications in HEP, LIGO, SDSS, Genomics

Korea

http://www.ivdgl.org/grid2003From “A Grid of One to a Grid of Many,” Miron Livny, UW-Madison, Keynote presentation, MIDnet conference, 2005.

22

National GridsMany countries have embraced grid computing and set-up grid computing infrastructure:

• UK e-Science grid• Grid-Ireland• NorduGrid• DutchGrid• POINIER grid (Poland)• ACI grid (France)• Japanese grid• etc, etc., …

23

UK e-Science Grid

24

Resource sharing and collaborative computing

• Grid computing is about collaborating and resource sharing as much as it is about high performance computing.

25

Virtual Organizations

Grid computing offers

potential of virtual organizations:– groups of people, both geographically and

organizationally distributed, working together on a problem, sharing computers AND other resources such as databases and experimental equipment.

• Crosses multiple administrative domains.

26

Applications

• Originally e-Science applications– Computational intensive

• Not necessarily one big problem but a problem that has to be solved repeatedly with different parameters.

– Data intensive.– Experimental collaborative projects

• Now also e-Business applications to improve business models and practices.

27(Based on a slide from HP)

Utility ComputingOne of Several Commercial Drivers

shared, traded resources

value

clusters

grid-enabled systems

programmable data center

virtual data center

Open VMS clusters, TruCluster, MC ServiceGuard

Tru64, HP-UX, Linux

switchfabriccompute storage

UDC

computing utility

or

GRID

today

• Utility computing• On-demand• Service-orientation• Virtualization

I. Foster

28

Grid Computing Software Infrastructure

29

Globus Project

• Open source software toolkit developed for grid computing.

• Roots in I-way experiment.• Work started in 1996. • Four versions developed to present time.• Reference implementations of grid computing

standards.• Defacto standard for grid computing.

30

Globus Toolkit:Recent History

• GT2 (2.4 released in 2002)– GRAM, MDS, GridFTP, GSI.

• GT3 (3.2 released mid-2004): redesign– OGSA (Open Grid Service Architecture)/OGSI (Open

Grid Services Infrastructure) based.– Introduced “Grid services” as an extension of web

services.– OGSI now abandoned.

• GT4 (release for April 2005): redesign– WSRF (Web service Resource Framework) based.– Grid standards merged with Web services.

31

Supercomputing 2003 Demonstration

• We* used Globus version 2.4 in a Supercomputing 2003 demo organized by the University of Melbourne.

• 21 countries involved, numerous sites.

* The Grid group at WCU.

33

A re-implementation based upon the Open Grid Service Architecture (OGSA) standard.

• We used version 3.2 for the Fall 2004 grid computing course.

• Underlying implementation of version 3.x used OGSI Open Grid Service Infrastructure), which was not embraced by the community.

Version 3

34

Version 4

• Released April 2005.

• OGSA kept but OGSI abandoned in favor of new implementation standards based around pure web services.(Version 3 used “extended” web services)

• To be used in this course, with other software.

35

Interconnections and Protocols

Focus now on:

• using standard Internet protocols and technology, i.e. HTTP, SOAP, web services, etc.,

36

Web Services-Based Grid Computing

• Grid Computing now strongly based upon web services.

• Large number of newly proposed grid computing standards:– WS-Resource Framework (WSRF)– WS-Addressing– etc., etc. …. .

37

Grid Computing Standards

ITCS 4010 Grid Computing, 2005, UNC-Charlotte, B. Wilkinson, slides 4a version 0.1.

38

Standards Bodies

Principal standards and other interested bodies are:

• W3C consortium (http://www.w3.org)

• Global Grid Forum (GGF)

• OASIS(Organization for the Advancement of Structured Information Standards)

• …..

http://www.w3.org/

39

In Web Services World• XML introduced (ratified) in 1998

• SOAP ratified in 2000

• Web services developed

• Subsequently, standards have been are continuing to be developed:– WSDL– WS-* where * refers to names of one of many

standards

40

• Originally own protocols were developed (e.g. GT2)

then• OGSA (Open Grid Services architecture)

standard, and a specification called OGSI (Open Grid Service Infrastructure) developed. Extended web service invented called a grid service to embody state and transience. (GGF) Implemented in GT3.

and• Now relies more directly upon developing

web service standards (GT 4)

1996-2002

2002-2004

2005 -

Grid computing software has gone through several development cycles:

41

Grid computing standards

• Figure from “An ‘Ecosystem’ of Grid Components”, 2004, Grid Research Integration Deployment and Support Center, http://www-unix.grids-

center.org/r6/ecosystem/ecology.php

42

Open Grid Services Architecture(OGSA)

Although OGSI vanished, OGSA continues …

43

OGSA

• Defines standard mechanisms for creating, naming, and discovering service instances.

• Addresses architectural issues relating to interoperable services for grid computing.

• Originally described in “The Physiology of the Grid” http://www.globus.org/research/papers/ogsa.pdf

44

WS-Resource Framework

• A specification developed by OASIS

• Specifies how to make web services stateful, and other feature

46From “The Globus Toolkit 4 Programmer’s Tutorial” by Borja Sotomayor.

47From “The Globus Toolkit 4 Programmer’s Tutorial” by Borja Sotomayor.

48

WS-* StandardsPrincipal web service standards adopted for grid computing:

• WSRF Framework collection of 5 specifications:– WS-ResourceProperties

• Specifies how resource properties are defined and accessed

– WS-ResourceLifetime• Specifies mechanisms to manage resource lifetimes

– WS-ServiceGroup• Specifies how to group services or WS-Resources together

– WS-BaseFaults• Specifies how to report faults

• WS-Notification– Collection of specifications that specifies how configure services are

notification producers or consumers

• WS-Addressing– Specifies how to address web services.– Provides a way to address a web service/resource pair

49

Grid Computing Software

Components of Globus 4.0

ITCS 4010 Grid Computing, 2005, UNC-Charlotte, B. Wilkinson, slides 4b version 0.1.

50

Globus Version 4

• A “toolkit” of services and packages for creating the basic grid computing infrastructure

• Higher level tools added to this infrastructure• Version 4 is web-services based• Some non-web services code exists from

earlier versions (legacy) or where not appropriate (for efficiency, etc.).

52

• Each part comprises a set of web services and/or non-web service components.

• Some built upon earlier versions of Globus.

Data Management

SecurityCommonRuntime

Execution Management

Information Services

Web Services

Components

Non-WS

Components

Pre-WSAuthenticationAuthorization

GridFTP

GridResource

Allocation Mgmt(Pre-WS GRAM)

Monitoring& Discovery

System(MDS2)

C CommonLibraries

GT2

WSAuthenticationAuthorization

ReliableFile

Transfer

OGSA-DAI[Tech Preview]

GridResource

Allocation Mgmt(WS GRAM)


System(MDS4)

Java WS Core

CommunityAuthorization

ServiceGT3

ReplicaLocationService

XIO

GT3

CredentialManagement

GT4

Python WS Core[contribution]

C WS Core

CommunitySchedulerFramework

[contribution]

DelegationService

GT4

Globus Open Source Grid Software

I Foster

54

Java Services in Apache AxisPlus GT Libraries and Handlers

YourJava

Service

YourPythonService

YourJava

Service RF

T

GR

AM

Del

egat

ion

Inde

x

Trig

ger

Arc

hive

r

pyGlobusWS Core

YourC

Service

C WS Core

RLS

Pre

-WS

MD

S

CA

S

Pre

-WS

GR

AM

Sim

pleC

A

MyP

roxy

OG

SA

-DA

I

GT

CP

Grid

FT

P

C Services using GT Libraries and Handlers

SERVER

CLIENT

InteroperableWS-I-compliant

SOAP messaging

YourJavaClient

YourC

Client

YourPythonClient

YourJavaClient

YourC

Client

YourPythonClient

YourJavaClient

YourC

Client

YourPythonClient

YourJavaClient

YourC

Client

YourPythonClient

X.509 credentials =common authentication

Python hosting, GT Libraries

Another view of GT4 Components

I Foster

55

GT Core

• Provides the ability to create services running inside the GT 4 container.

• Assignment 2 requires you to create a service inside GT 4 container and exercise it with a client.

Data Management




Web Services

Components

Non-WS

Components


GridFTP

GridResource



System(MDS2)

C CommonLibraries

GT2


ReliableFile

Transfer


GridResource



System(MDS4)

Java WS Core


ServiceGT3


XIO

GT3


GT4


C WS Core


[contribution]

DelegationService

GT4

Java WS Core

Used in assignment 2

57

CustomWeb

ServicesWS-Addressing, WSRF,

WS-Notification

CustomWSRF Web

Services

GT4WSRF Web

Services

WSDL, SOAP, WS-Security

User Applications

Reg

istr

yA

dmin

istr

atio

n

GT

4 C

onta

iner

GT4 Web Services Core

I Foster

58


Key component

GRAM (Grid Resource Allocation Manager)

• For submitting executable jobs• Used in Assignment 3 to submit and execute

jobs.• May interface to a local job scheduler• Local job scheduler used in assignment 4

Data Management




Web Services

Components

Non-WS

Components


GridFTP

GridResource



System(MDS2)

C CommonLibraries

GT2


ReliableFile

Transfer


GridResource



System(MDS4)

Java WS Core


ServiceGT3


XIO

GT3


GT4


C WS Core


[contribution]

DelegationService

GT4

GRAM (Grid Resource Allocation Manager)

Used in assignment 3

60

GRAMservices

GT4 Java Container

GRAMservices

Delegation

RFT FileTransfer

Transferrequest

GridFTPRemote storage element(s)

Localscheduler

Userjob

Compute element

GridFTP

sudo

GRAMadapter

FTPcontrol

Local jobcontrol

Delegate

FTP data

Cli

ent

Job

functions

Delegate

Service host(s) and compute element(s)

GT4 GRAM Structure:

Sun Grid Engine used in assignment 4

Data management components

I Foster

61

Security ComponentsAddresses the security requirements of grid computing. Three important factors are:

• Authorization– Process of deciding whether a particular identity can

access a particular resource

• Authentication– Process of deciding whether a particular identity is

who he says he is (applies to humans and systems)

• Delegation (somewhat specific to grid computing)– Process of giving authority to another identity

(usually a computer/process) to act on your behalf.

62

Security continued

• Security aspects complicated by the fact that virtual organization members and resources can be in different administrative domains.

Data Management




Web Services

Components

Non-WS

Components


GridFTP

GridResource



System(MDS2)

C CommonLibraries

GT2


ReliableFile

Transfer


GridResource



System(MDS4)

Java WS Core


ServiceGT3


XIO

GT3


GT4


C WS Core


[contribution]

DelegationService

GT4

Security

65

GT4’s Use of Security Standards

I Foster

66

GT4 Data Management

• Move large data to/from nodes• Replicate data for performance &

reliability• Locate data of interest• Provide access to different data sources

– File systems, parallel file systems, hierarchical storage (GridFTP)

– Databases (OGSA DAI)

Data Management




Web Services

Components

Non-WS

Components


GridFTP

GridResource



System(MDS2)

C CommonLibraries

GT2


ReliableFile

Transfer


GridResource



System(MDS4)

Java WS Core


ServiceGT3


XIO

GT3


GT4


C WS Core


[contribution]

DelegationService

GT4

GridFTP and Reliable File Transfer

68

GridFTP• Built on FTP using separation of data and

control channels• Provides features for

– Large data transfers– Secure transfers– Fast transfers– Reliable transfers– Third party transfers

• Not a web service– RTF (Reliable File Transfer) service provided WS-

level interface

69

Third party transfers

PI = FTP Protocol InterpreterDTP= FTP Data Channel Process

PI

DTP DTP

PI

PI PI

Client

Server Server

Control channels

Data channel

70

Performing a third-party transfer

1. Client establishes control channel with server2. Using control channel, client sets up transfer

parameters and requests data channel creation

3. Data channel established,4. Client sends transfer command over control

channel,5. Data transfer starts through data channel.

Either client or server can send.

71

Parallel transfers and striping

• Using multiple (virtual) connections for transfer– Same external network– Speed improvement possible, but limited by

network card

• Striping– a version of parallel transfers that can use

separate hardware interfaces– Implemented in GT 4.

72

GridFTP and RFT

WS ClientRFT service

(Java)

XIO based (C) XIO based (C)

Control channel

Data channel

GridFTP server GridFTP server

From Gridwise

Control channel

RequiresGSI proxy from client

73

GT 4 Replica Location Service

• Identify location of files via logical to physical name map

• Distributed indexing of names, fault tolerant update protocols

IndexIndex

I Foster

Data Management




Web Services

Components

Non-WS

Components


GridFTP

GridResource



System(MDS2)

C CommonLibraries

GT2


ReliableFile

Transfer


GridResource



System(MDS4)

Java WS Core


ServiceGT3


XIO

GT3


GT4


C WS Core


[contribution]

DelegationService

GT4

Monitoring and Discovery

75

Monitoring and Discovery• WSRF provides common mechanisms for

monitoring and discovering a service:• GT4 “aggregator” services within MDS:

– MDS-Index: collects state information from registered resources and makes it available as XML document

– MDS-Trigger: passes this information to an executable

– MDS-Archive: archives state information (awaiting implementation)

• Every GT 4 is discoverable

76

AcknowledgementSlides numbers marked with “I. Foster” have been

selected from presentations made by Ian Foster:• Enabling eScience: Grid Technologies Today &

TomorrowAmerican Association for the Advancement of Science Annual Meeting, Washington, DC, February 21 2005.

• Globus: Bridging the GapKeynote Talk, GlobusWORLD, Boston, Mass., February 8, 2005.

• The Grid: Reality, Technologies, ApplicationsDistinguished Lecture, McGill University, Montreal, Canada, January 21 2005.

used for educational purposes only.

77

AcknowledgementsSupport for this work was provided by:

National Science Foundation’s Course, Curriculum, and Laboratory Improvement program under grant # 0410667, “Introducing Grid Computing into the Undergraduate Curricula,”

University of North Carolina Office of the President through Award # P342A000189 “A Consortium to Promote Computational Science and High Performance Computing,”

University of North Carolina Office of the President through award # IR 04-04, “Fostering Undergraduate Research Partnerships through a Graphical User Environment for the North Carolina Computing Grid.”

The grid computing coursework development group gratefully acknowledges their support.

1 July 30, 2005 Grid Computing Principles Consortium for Computational Science and High Performance...

Documents

Transcript of 1 July 30, 2005 Grid Computing Principles Consortium for Computational Science and High Performance...