Grid Economics and Business Models: A Gridbus Perspective Rajkumar Buyya Grid and Distributed Systems (GRIDS) Laboratory Dept. of Computer Science and.

Grid Economics and Business Models: A Gridbus Perspective

Rajkumar BuyyaGrid and Distributed Systems (GRIDS) LaboratoryDept. of Computer Science and Software EngineeringThe University of MelbourneMelbourne, Australia

www.gridbus.org

WW GridGrid User Meet, Bengaluru, India

2

Outline

Introduction Utility Networks and Grid Computing

Global Grids and Challenges Security. Resource management, pricing and service

models Service Oriented Grids and Grid Economy

SOGA, Grid Market Directory, Grid Bank, Broker Grid Service Broker

Architecture, Design and Implementation Performance Evaluation: Experiments in Creation

and Deployment of Applications on Global Grids A Case Study in High Energy Physics

Summary

3

4 Essential Utilities and Delivery Networks

(1) Water

(2) Electricity

(3) Gas

(4) Telephone

4

(5) IT services as the fifth utility (water, electricity, gas, telephone, IT)

eScienceeBusiness

eGovernmenteHealth

MultilingualeEducation

…

5

Leading to Grid (computing) Paradigm:Cyberinfrastructure for sharing resources

•Inspired by Power Grid!

•* A service-oriented/utility computing paradigm that enables seamless

sharing of geographically distributed, autonomous resources (for PROFIT).

•* This was the original aim of building Internet although it ended up in

giving birth to email!

6

A Bird Eye View of World-Wide Grid Environment

Grid Resource Broker

Resource Broker

Application

Grid Information Service

Grid Resource Broker

databaseR2R3

RN

R1

R4

R5

R6

Grid Information Service

7

Various Types of Grid Services

Computational Services – CPU cycles SETI@Home, NASA IPG, TeraGrid, I-Grid,…

Data Services Data replication, management, secure

access--LHC Grid/Napster Application Services

Access to remote software/libraries and license management—NetSolve

Interaction Services eLearning, Virtual Tables, Group

Communication (Access Grid), Gaming Knowledge Services

The way knowledge is acquired and managed—data mining.

Utility Computing Services Towards a market-based Grid computing:

Leasing and delivering Grid services as ICT utilities.

Computational Grid

Data Grid

ASP Grid

Interaction Grid

Knowledge Grid

Utility Grid

8

Worldwide Grid Spending

After the year 2006, business popularity of Grid computing is expected to be accelerate exponentially:

Especially, the financial services and ERP services is expected to take major parts in the expense

Source: Insight Research Corp.

9

Grid Challenges

Security

Resource Allocation & Scheduling

Data locality

Network Management

System Management

Resource Discovery

Uniform Access

Computational Economy

Application Construction

10

Some Grid Initiatives Worldwide

Australia Nimrod-G Gridbus DISCWorld GrangeNet. APACGrid ARC eResearch

Brazil OurGrid, EasyGrid LNCC-Grid + many others

China ChinaGrid – Education CNGrid - application

Europe UK eScience EU Grids.. and many more...

India Garuda

Japan NAGERI

Korea...N*Grid

SingaporeNGP

USA Globus NASA IPG AccessGrid TeraGrid Cyberinfrasture

Industry Initiatives IBM On Demand

Computing HP Adaptive Computing Sun N1 Microsoft - .NET Oracle 10g Satyam – Grid Practice Infosys, Wipro, TCS StorageTek –Grid..

Public Forums Global Grid Forum Australian Grid Forum

Conferences: CCGrid Grid HPDC E-Science

http://www.gridcomputing.com

1.3 billion – 3 yrs

1 billion – 5 yrs

450million – 5 yrs

486million – 5 yrs

1.3 billion (Rs)

27 million

2? billion

120million – 5 yrs

11

Grid (Market) Participant Types and Application Category

commercialscientific

free trading

regulation

Publiccomputing

(SETI@Home, Alchemi, UD)

National provider(K*Grid, TeraGrid, Garuda/IndiaGrid, UKGrid, AusGrid)

Private enterprises(Satyam, IBM, Sun)

Application Category

SharingModel

12

mix-and-match (service)

Object-oriented

Internet/partial-P2P

Network enabled Solvers

Economic-based Utility / Service-Oriented

ComputingNimrod-G

13

The Gridbus Project @ Melbourne:Enable Leasing of ICT Services on

Demand

WWG

World Wide Grid!On Demand Utility

Computing

Gridbus

Distributed Data

14

The Gridbus Project @ GRIDS Lab, The University of Melbourne: Toolkit for Creating and Deploying e-Research Applications on Utility Grids

The Gridbus Project @ GRIDS Lab, The University of Melbourne: The Gridbus Project @ GRIDS Lab, The University of Melbourne: Toolkit for Creating and Deploying eToolkit for Creating and Deploying e--Research Applications on Utility GridsResearch Applications on Utility Grids

Gridbus

Distributed Data

http://www.gridbus.org

• Gridbus is a “open source” Grid R&D project with focus on Grid Economy, Utility Grids and Service Oriented Computing.

• Gridbus Middleware components include:– Alchemi: .NET-based Enterprise Grid

– Grid Market Directory and Web Services

– Grid Bank: Accounting and Transaction Management

– Visual Tools for Creation of Distributed Applications

– Grid Service Broker and Scheduling

– Workflow Management Engine

– GridSim Toolkit

– Libra: SLA-based Resource Allocation

15

Grid Economy: Methodology for Sustained Resourced Sharing and Managing Supply-and-Demand for Resources

16

New challenges of Grid Economy

Resource Owners How do I decide prices ? (economic models?) How do I specify them ? How do I translate price to resource allocation ? How do I enforce them ? How do I advertise & attract consumers ? How do I do accounting and handle payments? …..

Resource Consumers How do I decide expenses ? How do I express QoS requirements ? How do I trade between timeframe & cost ? How do I map jobs to resources to meet my QoS needs? …..

They need mechanisms and technologies for value expression, value translation, and value enforcement.

17

Grid Entities and Architecture

GSP site scheduler

accounting

Grid consumer

MarketMaker

GSP global scheduler

broker

Resource Provider

End usersPrivate enterprises

National providers

GSP site scheduler

Resource Provider

18

Grid Node N

A Reference Service-Oriented Architecture for Utility Grids

Grid Consumer

Pro

gra

mm

ing

En

viro

nm

ents

Grid Resource Broker

Grid Service Providers

Grid Explorer

Schedule Advisor

Trade Manager

Job ControlAgent

Deployment Agent

Trade Server

Resource Allocation

ResourceReservation

R1

Misc. services

Information Service

R2 Rm…

Pricing Algorithms

Accounting

Grid Node1

…

Grid Middleware Services

…

…

HealthMonitor

Grid Market Services

JobExec

Info ?

Secure

Trading

QoS

Storage

Sign-on

Grid Bank

Ap

pli

cati

on

s

Data Catalogue

19

Gridbus and Complementary Technologies – realizing Utility

Grid

AIXSolarisWindows Linux

.NET GridFabricSoftware

GridApplications

Core GridMiddleware

User-LevelMiddleware(Grid Tools)

GridBank

Grid Exchange & Federation

JVM

Grid Brokers:

X-Parameter Sweep Lang.

Gridbus Data Broker

MPI

Condor SGE TomcatPBS

Alchemi

Workflow

IRIX OSF1 Mac

Libra

Globus Unicore ……Grid

MarketDirectory

PDB

CDB

Worldwide Grid

GridFabricHardware

……

PortalsScience Commerce Engineering ……Collaboratories

……

Workflow Engine

Grid Storage Economy

Gri

d E

con

om

y NorduGrid XGrid

ExcellGrid

Nimrod-G

GRIDSIM

Gridscape

20

On Demand Assembly of Services: Putting Them All Together

ASP Catalogue

Grid Info Service

Grid Market Directory

GSP(Accounting Service)

GridbusGridBank

GSP(e.g., UofM)

PEGSP

(e.g., VPAC)

PE

GSP(e.g., IBM)

CPUorPE

Grid Service (GS)

(Globus)

Alchemi

GS

GTS

Cluster Scheduler

Job

8

GridResource Broker

2

Visual Application Composer

Application CodeExplore

data1

36

45

Resu

lts9 7

Results+

Cost Info

10

11

Bill

12Data Catalogue

21

Alchemi: .NET-based Enterprise Grid Platform & Web Services

InternetInternet

InternetInternet

Alchemi Worker Agents

Alchemi Manager

Alchemi Users

Web Services

Web Services

•SETI@Home like Model•General Purpose•Dedicated/Non-dedicate workers•Role-based Security•.NET and Web Services•C# Implementation•GridThread and Job Model Programming•Easy to setup and use• Widely in use!

22

Some Users of Alchemi

Tier Technologies, USALarge scale document processing using Alchemi framework

CSIRO, AustraliaNatural Resource Modeling

The Friedrich Miescher Institute (FMI) for Biomedical Research, SwitzerlandPatterns of transcription factors in mammalian genes

Satyam Computers Applied Research Laboratory, IndiaMicro-array data processing using Alchemi framework

The University of Sao Paulo, BrazilThe Alchemi Executor as a Windows Service

stochastix GmbH, GermanyAsynchronous Excel Tasks using ManagedXLL and Alchemi .Net Grid Computing framework.

Many users in Universities: See next for an example.

The Gridbus Grid Service Broker for Data Grid

Applications

Builds on the Nimrod-G Computational Grid Broker and

Computational Economy [Buyya, Abramson, Giddy, Monash

University, 1999-2001]And

Extends its notion for Data and Service Grids

24

Gridbus Broker Architecture

Grid Middleware

Gridbus Client Gridbus ClientGribus Client

Grid Info Server

Schedule Advisor

Trading Manager

Gridbus Farming Engine

RecordKeeper

Grid Explorer

GE GIS, NWSTM TS

RM & TS

Grid Dispatcher

RM: Local Resource Manager, TS: Trade Server

G

G

CU

Globus enabled node.A

L

Alchemi enabled node.

(Data Grid Scheduler)

DataCatalog

DataNode

Unicore enabled node.

$

$

$

App, T, $, Opt

(Bag of Tasks Applications)

26

Gridbus Services for eScience applications

Application Development Environment: XML-based language for composition of task farming

(legacy) applications as parameter sweep applications. Task Farming APIs for new applications. Web APIs (e.g., Portlets) for Grid portal development. Threads-based Programming Interface Workflow interface and Gridbus-enabled workflow

engine. Resource Allocation and Scheduling

Dynamic discovery of optional computational and data nodes that meet user QoS requirements.

Hide Low-Level Grid Middleware interfaces Globus, Alchemi, Unicore, NorduGrid, XGrid, etc.

27

Drug DesignMade Easy!

Click Here for Demo

Economy-based Data Grid Scheduling

High Energy Physics as eScience Application Case

Study

CLICK HERE TO SKIP IF RUNNING OUT of TIME

29

Case Study: High Energy Physics and Data Grid The Belle Experiment

KEK B-Factory, Japan Investigating fundamental

violation of symmetry in nature (Charge Parity) which may help explain the universal matter – antimatter imbalance.

Collaboration 400 people, 50 institutes

100’s TB data currently

30

Australian Belle Data Grid Testbed

Grid Service Broker

Replica Catalog

AARNET

NWS NameServer

VirtualOrganization

Analysis Request

Analysis Results

CertificateAuthority

NWSSensor

GridFTPGRIS

GlobusGatekeeper

Dual Intel Xeon 2.8 Ghz, 2 GB RAM

NWSSensor

GridFTPGRIS

GlobusGatekeeper

Dual Intel Xeon 2.8 Ghz, 2 GB RAM

NWSSensor

GridFTPGRIS

GlobusGatekeeper

Dual Intel Xeon 2.8 Ghz, 2 GB RAM

GRIDS Lab, University of Melbourne

Dept. of Physics,University of Sydney

ANU, Canberra

Dept. of Computer Science, University of Adelaide

NWSSensor

GridFTPGRIS

GlobusGatekeeper

Intel Pentium 2.0 Ghz, 512 MB RAM

Dept. of Physics,University of Melbourne

NWSSensor

GridFTPGRIS

GlobusGatekeeper

Dual Intel Xeon 2.8 Ghz, 2 GB RAM

VPACMelbourne

31

Case Study: Event Simulation and Analysis

B0->D*+D*-Ks

• Simulation and Analysis Package - Belle Analysis Software Framework (BASF)• Experiment in 2 parts – Generation of Simulated Data and Analysis of the distributed data

Analyzed 100 data files (30MB each) were distributed among the five nodes

32

Belle Data Grid (GSP CPU Service Price: G$/sec)

Grid Service Broker

Replica Catalog

AARNET

NWS NameServer

VirtualOrganization

Analysis Request

Analysis Results

CertificateAuthority

NWSSensor

GridFTPGRIS

GlobusGatekeeper

Dual Intel Xeon 2.8 Ghz, 2 GB RAM

NWSSensor

GridFTPGRIS

GlobusGatekeeper

Dual Intel Xeon 2.8 Ghz, 2 GB RAM

NWSSensor

GridFTPGRIS

GlobusGatekeeper

Dual Intel Xeon 2.8 Ghz, 2 GB RAM

GRIDS Lab, University of Melbourne

Dept. of Physics,University of Sydney

ANU, Canberra

Dept. of Computer Science, University of Adelaide

NWSSensor

GridFTPGRIS

GlobusGatekeeper

Intel Pentium 2.0 Ghz, 512 MB RAM

Dept. of Physics,University of Melbourne

NWSSensor

GridFTPGRIS

GlobusGatekeeper

Dual Intel Xeon 2.8 Ghz, 2 GB RAM

NA

G$4

G$4

Datanode

G$6VPAC

MelbourneG$2

33

Belle Data Grid (Bandwidth Price: G$/MB)

Grid Service Broker

Replica Catalog

AARNET

NWS NameServer

VirtualOrganization

Analysis Request

Analysis Results

CertificateAuthority

NWSSensor

GridFTPGRIS

GlobusGatekeeper

Dual Intel Xeon 2.8 Ghz, 2 GB RAM

NWSSensor

GridFTPGRIS

GlobusGatekeeper

Dual Intel Xeon 2.8 Ghz, 2 GB RAM

NWSSensor

GridFTPGRIS

GlobusGatekeeper

Dual Intel Xeon 2.8 Ghz, 2 GB RAM

GRIDS Lab, University of Melbourne

Dept. of Physics,University of Sydney

ANU, Canberra

Dept. of Computer Science, University of Adelaide

NWSSensor

GridFTPGRIS

GlobusGatekeeper

Intel Pentium 2.0 Ghz, 512 MB RAM

Dept. of Physics,University of Melbourne

NWSSensor

GridFTPGRIS

GlobusGatekeeper

Dual Intel Xeon 2.8 Ghz, 2 GB RAM

NA

G$4

G$4

Datanode

G$6VPAC

MelbourneG$2

34

31

38

31

30

3336

32

35

Deploying Application Scenario

A data grid scenario with 100 jobs and each accessing remote data of ~30MB

Deadline: 3hrs. Budget: G$ 60K Scheduling Optimisation Scenario:

Minimise Time Minimise Cost

Results:

SUMMARY OF EVALUATION RESULTS

Scheduling strategy Total Time Taken (mins.)

Compute Cost (G$)

Data Cost (G$)

Total Cost (G$)

Cost Minimization 71.07 26865 7560 34425 Time Minimization 48.5 50938 7452 58390

36

Time Minimization in Data Grids

0

10

20

30

40

50

60

70

80

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42

Time (in mins.)

Nu

mb

er

of

job

s c

om

ple

ted

fleagle.ph.unimelb.edu.au belle.anu.edu.au belle.physics.usyd.edu.au brecca-2.vpac.org

37

Results : Cost Minimization in Data Grids

0

10

20

30

40

50

60

70

80

90

100

1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31 33 35 37 39 41 43 45 47 49 51 53 55 57 59 61 63

Time(in mins.)

Nu

mb

er o

f jo

bs

com

ple

ted

fleagle.ph.unimelb.edu.au belle.anu.edu.au belle.physics.usyd.edu.au brecca-2.vpac.org

38

SUMMARY OF EVALUATION RESULTS

Scheduling strategy Total Time Taken (mins.)

Compute Cost (G$)

Data Cost (G$)

Total Cost (G$)

Cost Minimization 71.07 26865 7560 34425 Time Minimization 48.5 50938 7452 58390

Observation

Organization 

Node details Cost (in G$/CPU-sec)

Total Jobs Executed

Time Cost

CS,UniMelb belle.cs.mu.oz.au4 CPU, 2GB RAM, 40 GB HD, Linux

N.A. (Not used as a compute resource)

-- --

Physics, UniMelb fleagle.ph.unimelb.edu.au1 CPU, 512 MB RAM, 40 GB HD, Linux

2 3 94

CS, University of Adelaide

belle.cs.adelaide.edu.au4 CPU (only 1 available) , 2GB RAM, 40 GB HD, Linux

N.A. (Not used as a compute resource)

-- --

ANU, Canberra belle.anu.edu.au4 CPU, 2GB RAM, 40 GB HD, Linux

4 2 2

Dept of Physics, USyd

belle.physics.usyd.edu.au4 CPU (only 1 available), 2GB RAM, 40 GB HD, Linux

4 72 2

VPAC, Melbourne

brecca-2.vpac.org180 node cluster (only head node used), Linux

6 23 2

39

Grid and Gridbus Technologies for Various Grid (Market) Types

commercialscientific

free trading

regulation

Publiccomputin

g(Alchemi)

National provider(Globus, Gridbus,..)

Private enterprises

(Libra, Gridbus, Globus)

Application Category

SharingModel

41

Summary and Conclusion

Grids exploit synergies that result from cooperation of autonomous entities:

Resource sharing, dynamic provisioning, and aggregation at global level.

Grid Economy provides incentive needed for sustained cooperation.

Grid Network has potential to serve as Cyberinfrastructure for Utility Computing

Grids offer enormous opportunities for realizing eScience and eBusiness at global level.

42

Any Questions ?

Gridbus Project - http://www.gridbus.org

43

Thanks for your attention!

The Gridbus Cooperation!http://www.gridbus.com

Backup Slides

47

Deadline (D) and Budget (B) Constrained Scheduling Algorithms

Algorithm

Execution Time (D)

Execution Cost (B)

Compute Grid

Data Grid

Cost Opt Limited by D

Minimize Yes Yes

Cost-Time Opt

Minimize if possible

Minimize Yes

Time Opt Minimize Limited by B

Yes Yes

Conservative-Time Opt

Minimize Limited by B, jobs have guaranteed minimum budget

Yes