Computational Grid Portals Marlon Pierce Community Grids Lab Indiana University.

12
Computational Grid Computational Grid Portals Portals Marlon Pierce Marlon Pierce Community Grids Lab Community Grids Lab Indiana University Indiana University

Transcript of Computational Grid Portals Marlon Pierce Community Grids Lab Indiana University.

Page 1: Computational Grid Portals Marlon Pierce Community Grids Lab Indiana University.

Computational Grid Computational Grid PortalsPortals

Marlon PierceMarlon Pierce

Community Grids LabCommunity Grids Lab

Indiana UniversityIndiana University

Page 2: Computational Grid Portals Marlon Pierce Community Grids Lab Indiana University.

Open Grid Computing Open Grid Computing Environment CollaboratoryEnvironment Collaboratory

Funded by NSF’s National Middleware InitiativeFunded by NSF’s National Middleware Initiative University of ChicagoUniversity of Chicago

• Gregor von Laszewski Gregor von Laszewski University of Illinois/NCSAUniversity of Illinois/NCSA

• Jay AlamedaJay Alameda• Joe FutrelleJoe Futrelle

Indiana University/Community Grids Lab and CSIndiana University/Community Grids Lab and CS• Marlon PierceMarlon Pierce• Geoffrey Fox Geoffrey Fox • Dennis GannonDennis Gannon• Beth PlaleBeth Plale

University of MichiganUniversity of Michigan• Charles SeveranceCharles Severance• Joseph HardinJoseph Hardin

University of Texas/TACCUniversity of Texas/TACC• Mary ThomasMary Thomas• Jay BoisseauJay Boisseau

Page 3: Computational Grid Portals Marlon Pierce Community Grids Lab Indiana University.

What Are Grid Portals?What Are Grid Portals? Computing portals provide ubiquitous, browser-based Computing portals provide ubiquitous, browser-based

access to grid resources.access to grid resources.• No special client software or platform neededNo special client software or platform needed• Access information in visually intuitive form Access information in visually intuitive form

Provide services to support user interactionsProvide services to support user interactions• Job archivingJob archivingportal metadata management servicesportal metadata management services

Combine core grid services into custom servicesCombine core grid services into custom services• Launch multistage jobs with dependenciesLaunch multistage jobs with dependencies• Couple execution, file transfer, visualization/analysisCouple execution, file transfer, visualization/analysis

Many, many such projectsMany, many such projects• Concurrency and Computation: Practice and Experience’sConcurrency and Computation: Practice and Experience’s

special issue described more than two dozen in 2001.special issue described more than two dozen in 2001. GCE Research Group of the GGF is the community forum.GCE Research Group of the GGF is the community forum.

• Thomas, Gannon, and Fox are chairs.Thomas, Gannon, and Fox are chairs.

Page 4: Computational Grid Portals Marlon Pierce Community Grids Lab Indiana University.

What Are the Problems?What Are the Problems? Problems are always the same:Problems are always the same:

• How do we share portal services?How do we share portal services?• How do we reuse components between projects and How do we reuse components between projects and

groups?groups? Can we provide a standard abstraction for Can we provide a standard abstraction for

portal services and interfaces?portal services and interfaces? Can we provide an architecture that allows Can we provide an architecture that allows

services and user interface components to be services and user interface components to be added in a standard way?added in a standard way?

Need to shorten the “standard service” Need to shorten the “standard service” deployment phase so that we can concentrate deployment phase so that we can concentrate on harder problems, specific sophisticated on harder problems, specific sophisticated servicesservices• Fusion Grid needs very interactive, visual interface Fusion Grid needs very interactive, visual interface

for setting up problemsfor setting up problems• Need to be able to deploy standard components like Need to be able to deploy standard components like

MyProxy, GridFTP, etc interfaces quicklyMyProxy, GridFTP, etc interfaces quickly

Page 5: Computational Grid Portals Marlon Pierce Community Grids Lab Indiana University.

General Portal ArchitecturesGeneral Portal Architectures

Page 6: Computational Grid Portals Marlon Pierce Community Grids Lab Indiana University.

Portal Service ComponentsPortal Service ComponentsSciDAC Fusion Grid ExampleSciDAC Fusion Grid Example

GridAuth

GRAMGridFTP

GRIS/GIIS

TRANSPMDSPlus

SRBIDL

Services

WSDL WSDL

GPIR

WSDL

WSDL

SRBClient

Portlet Container

GridPort

PortalLogin

Java COG

TRANSPSubmit

WSDL

GPIR

COG/GP

ProxyManager

Java COG

LDAPBrowser

WSDL

FileManager

WSDL

IDL Clients

FG HardwareFG Applications

Page 7: Computational Grid Portals Marlon Pierce Community Grids Lab Indiana University.

SERVOGrid Codes, RelationshipsSERVOGrid Codes, Relationships

Elastic DislocationPattern Recognizers

Fault Model BEM

Viscoelastic Layered BEM

Viscoelastic FEMElastic Dislocation Inversion

Page 8: Computational Grid Portals Marlon Pierce Community Grids Lab Indiana University.

Problems: Data Access and Problems: Data Access and Sharing, Code IntegrationSharing, Code Integration

Codes all use custom text formats for Codes all use custom text formats for describing input and output.describing input and output.

Input and output data often combined Input and output data often combined with code-specific information.with code-specific information.• Number of iterations, array sizes, etc.Number of iterations, array sizes, etc.

Data files often created by hand from Data files often created by hand from journals, online repositoriesjournals, online repositories• Online repositories themselves use differing Online repositories themselves use differing

formatsformats Challenges are to develop common data Challenges are to develop common data

formats, access services, and client query formats, access services, and client query tools.tools.

Page 9: Computational Grid Portals Marlon Pierce Community Grids Lab Indiana University.

JSP + Client Stubs

DB Service 1

JDBC

DB

Job Sub/Mon And FileServices

Operating andQueuing Systems

Browser Interface

Viz Service

Host 1 Host 2 Host 3

RIVA

Page 10: Computational Grid Portals Marlon Pierce Community Grids Lab Indiana University.

Web Services for Data Access and Web Services for Data Access and Computing Service InvocationComputing Service Invocation

Service-oriented architectureService-oriented architecture Web services: Web services:

• WSDLWSDL: Interface definition language, describes your service : Interface definition language, describes your service ““GeoFEST may be invoked with these input types”GeoFEST may be invoked with these input types”

• SOAPSOAP: Transport envelope for remote procedure calls/messages: Transport envelope for remote procedure calls/messages ““Invoke GeoFEST with this set of input”Invoke GeoFEST with this set of input”

Together, WSDL and SOAP are useful for manipulating, returning Together, WSDL and SOAP are useful for manipulating, returning XML data valuesXML data values

WSDL and SOAP are platform independentWSDL and SOAP are platform independent• C/C++, Java, Python, Perl bindings.C/C++, Java, Python, Perl bindings.

So you can build a variety of clients that use the same backend So you can build a variety of clients that use the same backend servicesservices• Browsers and more sophisticated workbench GUIs.Browsers and more sophisticated workbench GUIs.

Turn the problem on its head: embed service calls in legacy codeTurn the problem on its head: embed service calls in legacy code• Better data access methodsBetter data access methods• I want to use your mesh generatorI want to use your mesh generator

Page 11: Computational Grid Portals Marlon Pierce Community Grids Lab Indiana University.

GML Schemas as Data Models for GML Schemas as Data Models for ServicesServices

Fault and GPS Schemas are based on GML-Fault and GPS Schemas are based on GML-Feature object.Feature object.

Seismicity Schema is based on GML-Observation Seismicity Schema is based on GML-Observation object.object.

Working schema available from Working schema available from http://http://grids.ucs.indiana.edu/~gaydin/schemasgrids.ucs.indiana.edu/~gaydin/schemas//

Page 12: Computational Grid Portals Marlon Pierce Community Grids Lab Indiana University.

Grid Testbed for Portal DemoGrid Testbed for Portal Demo complexity.ucs.indiana.educomplexity.ucs.indiana.edu

• Solaris Sunfire web serverSolaris Sunfire web server• Web server for portalWeb server for portal

Danube.ucs.indiana.eduDanube.ucs.indiana.edu• Duel processor linux serverDuel processor linux server• Runs AKRIA,GeoFEST, GMTRuns AKRIA,GeoFEST, GMT

Grids.ucs.indiana.edu: Grids.ucs.indiana.edu: • Sun Ultra 60Sun Ultra 60• Runs DislocRuns Disloc

Infogroup.usc.eduInfogroup.usc.edu• Linux boxLinux box• Runs Mysql fault databaseRuns Mysql fault database

Jabba.jpl.nasa.gov:Jabba.jpl.nasa.gov:• SGI Origin(?)SGI Origin(?)• Runs RIVA for moviesRuns RIVA for movies