Recent Advances in Grid Computing and Business Models: A Gridbus Perspective Rajkumar Buyya Grid and Distributed Systems (GRIDS) Laboratory Dept. of Computer.

Recent Advances in Grid Computing 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 Business Symposium 2005, Seoul, Korea

2

Outline

Introduction Utility Networks and Grid Computing

Global Grids and Challenges Grid Initiatives

World-wide with Australia and India Perspective Introduction to Gridbus Project and Grid Economy 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 Economy-based Scheduling in Data Grids

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

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

Grid Challenges

Security

Resource Allocation & Scheduling

Data locality

Network Management

System Management

Resource Discovery

Uniform Access

Computational Economy

Application Construction

8

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 I-Grid

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

10

Grid Computing in Australia(Courtesy: Jihyoun Park, SNU Visitor to

Melbourne)

AcademiaGovernment

Collaboration

Industry

11

Academic activities

1 University laboratories for Grid computing- Uni. of Melbourne(GRIDS lab): Gridbus (GridSim, GMD, GridBank, Alchemi, ..), Master of Engineering in Distributed Computing

- Monash Uni.: GriddlsS (Legacy SW to the computational grid), Nimrod-G - Australian national Uni. (Internet Futures Group)- Sydney Uni.(ViSLAB): high performance visualization &computing - Uni. of Adelaide (DHPC Group): DISCWorld - Queensland Uni. of Technology (PLAS): G2 (.NET based)

2 Grid Infrastructure ProjectsAPACGrid, National Neurosciece Facility, Australian Virtual Observatory, several state level facilities (VPAC, TPAC, SAPAC, QPSF, IVEC)

3 Grid Applications * Asia Pacific Bioinformatics Network/ Virtual Drug Design: Molecular

Modeling for Drug Design on P2P Grid/ HEPGrid: High Energy Physics and the Grid Network/ Access Grid/Australian Computational Earth Systems Simulator/.

* Recently 30 more applications are funded as part of ARC e-Research * Govt. has formed “National e-Research Coordination Committee”.

12

Grid Computing in India

AcademiaGovernment Collabor

ation

Industry(majority focus onGrid integration)

13

Grid Computing in India: Academic and Industrial Activities

Academic and Government Initiatives: TIFR, IITM, Anna University, IITD, UoH, etc. C-DAC’s Garuda – Ministry of IT

Software Companies in India: Top 4 Indian IT Companies: Satyam, Infosys, TCS (Tata

Consultancy Service), and Wipro. Oracle 10g, IBM, HP, Sun ertc. have a large Grid

development centers in Bangalore, India. Satyam is leading the pack in Grid Business push:

Grid Practice Centre with top management support. Singned MoU with Melbourne University and

extensively using Gridbus in powering applications. Also contributing the development of Gridbus

technologies (e.g., Alchemi) – SEI CMM Level 5 principles.

Application Verticals: Manufacturing, Security, Life Sciences, Finance

14

15

Australian and Indian Grid Efforts Compared

AcademiaGovernment

Collaboration

Industry

Australia

AcademiaGovernment Collabor

ation

Industry(majority focus onGrid integration)

AcademiaGovernment Collabor

ation

Industry(majority focus onGrid integration)

IndiaKorea: Is it like Australia or India?

16

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

Demand

WWG

World Wide Grid!On Demand Utility

Computing

Gridbus

Distributed Data

17

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

A multi-institutional “Open Source” R&D Project with focus on: Architecture, Specification, and Open Source Reference Implementation. Service-Oriented Grid, Utility Computing & Distributed Data and Computation

Economy Scaling from Desktops, Clusters, Cluster Federation, Enterprise Grids to Global Grids.

Alchemi: Harnessing .NET/Windows-based Resources Grid Market Directory and Web Services Grid Bank: Accounting and Transaction Management Visual Tools for Creation of Distributed Applications Workflow Composition and Deployment Services Data Grid Brokering and Grid Economy Services Data Replication Strategies GridSim Toolkit: Enhanced to support Data Grid, Reservation, etc. Libra: SLA-based Allocation of Cluster Resources Coupling of Clusters and Computational Economy WWG: Global Data Intensive Grid Testbed Application Enabler Projects:

High-Energy Physics , Astronomy, Brain Activity Analysis – Osaka U., Natural Language Processing, Portfolio Analysis – Spain, BioGrid - WEHI (via APACGrid), SensorGrid (NICTA), Medical Imaging (HFI)

Supported by:

18

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

20

Grid Entities and Architecture

GSP site scheduler

accounting

Grid consumer

MarketMaker

GSP global scheduler

broker

Resource owners

End usersPrivate enterprises

National providers

GSP site scheduler

Resource owners

21

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

22

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

23

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!

24

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.

25

On Demand Assembly of Services: Putting Them All Together

Data Source

(Instruments/distributed sources)

Data Replicator(GDMP) ASP Catalogue

Grid Info Service

Grid Market Directory

GSP(Accounting Service)

GridbusGridBank

Data

GSP(e.g., UofM)

PEGSP

(e.g., VPAC)

PE

GSP(e.g., IBM)

CPUorPE

Grid Service (GS)

(Globus)

Alchemi

GS

GTS

Cluster Scheduler

Grid Service Provider (GSP)

(e.g., CERN)

PECluster 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

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

27

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)

29

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.

30

Figure 3 : Logging into the portal.

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

32

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

33

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

34

Resources Used and their Service Price

Organization 

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

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

Broker host, Data host, NWS server

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

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

Replica Catalog host, Data host, Compute resource, NWS sensor

2

CS, University of Adelaide

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

Data host, NWS sensor

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

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

Data host, Compute resource, NWS sensor

4

Dept of Physics, USyd

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

Data host, Compute resource, NWS sensor

4

VPAC, Melbourne

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

Compute resource,NWS sensor

6

35

Network Cost (in Grid $/Currency!)

NETWORK COSTS BETWEEN THE DATA HOSTS AND THE COMPUTE RESOURCES

(IN G$ PER MB) Data Node

Compute Node ANU UniMelb

Physics Sydney Physics

VPAC

ANU 0 34.0 31.0 38.0 Adelaide CS 34.0 36.0 31.0 33.0 UniMelb Physics 40.0 0 32.0 39.0 UniMelb CS 36.0 30.0 33.0 37.0 Sydney Physics 35.0 33.0 0 37.0

36

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

37

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

39

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.

40

Any Questions ?

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

41

Thanks for your attention!

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

Backup Slides

45

What do Grids aim for and how to support them.

Grids aim at exploiting synergies that result from cooperation of autonomous distributed entities. Synergies include:

Resource sharing “On-demand” Virtual Enterprises creation Aggregation of resources on demand.

For this cooperation to be sustainable, participants needs to have (economic) incentive.

Therefore, “incentive” mechanisms should be considered as one of key design parameters of Grid computing.

46

Grid Market (Participant) Types and Application Category

commercialscientific

free trading

regulation

Publiccomputin

g

National provider

Private enterprises

Application Category

SharingModel

47

Appropriate Market Model for different market types

strongweak

high

low

Variable price

auction

Posted price

oligopoly

Commodity market

Demand elasticity

Willingness to Pay

49

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