Nimrod/G GRID Resource Broker and Computational Economy David Abramson, Rajkumar Buyya, Jon Giddy School of Computer Science and Software Engineering Monash.

Nimrod/G GRID Resource Broker and Computational Economy

David Abramson, Rajkumar Buyya, Jon GiddySchool of Computer Science and Software Engineering

Monash University, Melbourne, Australia

Email: {davida, rajkumar, jon}@csse.monash.edu.au

Computing Power (HPC) DriversComputing Power (HPC) Drivers

Solving grand challenge applications using Solving grand challenge applications using computer computer modelingmodeling, , simulationsimulation and and analysisanalysis

Life SciencesLife Sciences

CAD/CAMCAD/CAM

AerospaceAerospace

Military ApplicationsDigital BiologyDigital Biology Military ApplicationsMilitary Applications

E-commerce/anything

2100

2100 2100 2100 2100

2100 2100 2100 2100

Desktop(Single Processor?)

SMPs or SuperCom

puters

LocalCluster

GlobalCluster/Grid

PERFORMANCE

Computing Platforms EvolutionComputing Platforms Evolution Breaking Administrative BarriersBreaking Administrative Barriers

Inter PlanetCluster/Grid ??

IndividualGroupDepartmentCampusStateNationalGlobeInter PlanetUniverse

Administrative Barriers

EnterpriseCluster/Grid

?

Global Computational GridsGlobal Computational Grids

Domain 2

Domain 1

Grid Resource Management: Grid Resource Management: Challenging IssuesChallenging Issues

Ack.: globus..

•Authentication (once)

•Specify simulation

(code, resources, etc.)

•Discover resources

•Negotiate authorization,

acceptable use, Cost, etc.

•Acquire resources

•Schedule Jobs

•Initiate computation

•Steer computation

•Access remote data-sets

•Collaborate on results

•Account for usage

Grid ComponentsGrid Components

GridFabricNetworked Resources across Organisations

Computers Clusters Data Sources Scientific InstrumentsStorage Systems

Local Resource Managers

Operating Systems Queuing Systems TCP/IP & UDP

…

Libraries & App Kernels …

Distributed Resources Coupling Services

Comm. Sign on & Security Information … QoSProcess Data Access

Development Environments and Tools

Languages Libraries Debuggers … Web toolsResource BrokersMonitoring

Applications and Portals

Prob. Solving Env.Scientific …CollaborationEngineering Web enabled Apps

GridApps.

GridMiddleware

GridTools

Grid Node N

Grid Node 2

Computational Market Model for Grid Resource ManagementComputational Market Model for Grid Resource Management

Grid User

Application

Grid Resource Broker

Grid Resource/Control Domains

Grid Explorer

Schedule Advisor

Trade Manager

Job ControlAgent

Deployment Agent

Trade Server

Resource Allocation

ResourceReservation

R1

Other services

Grid Information Server(s)

R2 Rm…

Charging Alg.

Accounting

Grid Node1

…

Trading

Grid Middleware

…

Info ?

…

Jobs

HealthMonitor

A global scheduler for managing and steering task A global scheduler for managing and steering task farming (parametric simulation) applications on farming (parametric simulation) applications on computational grid based on computational grid based on deadlinedeadline and and computational economycomputational economy. .

Key FeaturesKey Features– A single window to manage & control experimentA single window to manage & control experiment– Resource DiscoveryResource Discovery– Trade for ResourcesTrade for Resources– SchedulingScheduling– Steering & data managementSteering & data management

It allows to study the behaviour of some of the output It allows to study the behaviour of some of the output variables against a range of different input scenarios.variables against a range of different input scenarios.

What is Nimrod/G ?What is Nimrod/G ?

Nimrod/G Grid Resource Broker ArchitectureNimrod/G Grid Resource Broker Architecture

Grid Middleware

Nimrod/G Client Nimrod/G ClientNimrod/G Client

Grid Information Server(s)

Schedule Advisor

Trading Manager

Nimrod/G Engine

Grid Bookkeeper

Grid Explorer

GE GISTM TS

RM & TSRM & TS

Grid Dispatcher

RM: Local Resource Manager, TS: Trade Server

RM & TS

Globus,Legion, Condor-g,, Ninf,etc.

G

L

NG

Globus enabled node.Ninf enabled node.

CL

Condor enabled node.

Nimrod/G InteractionsNimrod/G Interactions

Grid Infoservers

Resource location

QueuingSystem

Processserver

Resource allocation

(local)User

process

File accessI/Oserver

Gatekeeper node

JobWrapper

Computational node

Dispatcher

Root node

Scheduler

Prmtc..Engine

Trade Server

A Nimrod/G ClientA Nimrod/G Client

CostCostDeadlineDeadline

Legion hosts

Globus Hosts

Bezek is in both Globus and Legion Domains

Arlington

Alexandria

Richmond

HamptonNorfolk

Virginia BeachChesapeakePortsmouth

Newport News

Roanoke

Ap p om a toxRiver

Ja m esRiver

Shena nd oa hRiver

Ra p p a ha nnoc kRiver

Potom a cRiver

VIRGINIA77

81

64

64

66

85

Change deadline/budget + Monitor activitiesChange deadline/budget + Monitor activities

Adaptive Scheduling algorithms ...Adaptive Scheduling algorithms ...

LocateLocate

MachinesMachines

DistributeDistribute

JobsJobs

EstablishEstablish

RatesRates

Meet requirements ?Meet requirements ?

Deadlines and BudgetDeadlines and Budget

Re-distributeRe-distribute

JobsJobs

LocateLocate

moremore

MachinesMachines

Adaptive SchedulingAlgorithms

Execution Time (not beyond deadline)

Execution Cost (not beyond budget)

Time Minimisation Minimise Limited by budgetCost Minimisation Limited by deadline MinimiseNone Minimisation Limited by deadline Limited by budget

Graph 5 - GUSTO Usage for 10 Hour Deadline

0

5

10

15

20

25

30

35

0 2.5 5 7.5 10 12.5 15 17.5 20

Time

No

Pro

cess

es

5 CUs

10 CUs

15 CUs

20 CUs

50 CUs

10 Cost Units

50 Cost Units

20 Cost Units

5 Cost Units

15 Cost Units

Graph 4 - GUSTO Usage for 15 Hour Deadline

0

2

4

6

8

10

12

14

16

18

20

0 2.5 5 7.5 10 12.5 15 17.5 20

Time

Ave

rag

e N

o P

roc

es

so

rs

5 CUs

10 CUs

15 CUs

20 CUs

50 CUs

5 Cost Units

50 Cost Units

15 Cost Units

10 Cost Units

Graph 3 - GUSTO Usage for 20 Hour Deadline

0

2

4

6

8

10

12

14

16

18

20

0 2.5 5 7.5 10 12.5 15 17.5 20

Time

Ave

rag

e N

o P

roce

sso

rs

5 CUs

10 CUs

15 CUs

20 CUs

50 CUs

5 Cost Units

10 Cost Units

Graph 2 - GUSTO Usage for Ionization Chamber Study

0

10

20

30

40

50

60

70

80

0 2.5 5 7.5 10 12.5 15 17.5 20

Time

Ave

rag

eN

o. P

roce

sso

rs20 Hour deadline15 hour deadline10 hour deadline

Nimrod/O

DeclarativePlanFile Simulated Annealing

Divide & Conquer

Simplex

P-BFGS

JobControl

Function Requests

FunctionValues

Nimrodor

Clustor

SupercomputerCluster

Jobs

Results

NimCache

Function Requests

FunctionValues

Nimrod/O: Automatic Design Nimrod/O: Automatic Design OptimizationOptimizationSearch parameter space rather than exploring all optionsSearch parameter space rather than exploring all options

NimCacheNimCache Nimrod/GNimrod/G

Active Sheets Active Sheets - Excell cell func()s execution on the Grid - Excell cell func()s execution on the Grid

Related Work & Further Info.Related Work & Further Info. AppLeS (UC. San Diego)AppLeS (UC. San Diego)

– application level scheduling templates case-by-case for different Apps, soon application level scheduling templates case-by-case for different Apps, soon PST.PST.

NetSolve (UTK/ORNL) -- API for creating farmsNetSolve (UTK/ORNL) -- API for creating farms

SETI @ Home, Distributed.net, ….SETI @ Home, Distributed.net, ….

Millennium (UC. Berkeley)Millennium (UC. Berkeley)– remote execution environment on clusters and supports remote execution environment on clusters and supports

computational economycomputational economy

CODINE/GRD (Genias/Gridware)CODINE/GRD (Genias/Gridware)– meets deadline by dominating over others share. meets deadline by dominating over others share.

Mariposa- Distributed Database system (UC, Berkeley)Mariposa- Distributed Database system (UC, Berkeley)– query with budget, creates sub-query & dividesbudget, trades with query with budget, creates sub-query & dividesbudget, trades with

(remote) servers(remote) servers

More Info -- More Info -- www.csse.monash.edu.au/~davida/nimrod.htmlwww.csse.monash.edu.au/~davida/nimrod.html