GridLab Applications

Gabrielle Allen, Zhou Lei, Archit Kulshrestha

Center for Computation & Technology, LSU

Modeling Black Holes• Einstein’s equations of General Relativity: Gab = 0

• One of the most complex set of equations in physics– Mixed hyperbolic-elliptic initial value problem– 3 Dimensional, multiple scales– 12 or more equations, with thousands of terms

• Require diverse distributed collaborations & many large scale simulations:– Parallelization, optimization– Parallel IO, data management, viz– Checkpointing, steering, interaction

t=100

t=0

Cactuscode.org

Application Scenarios• Intelligent Automatic Parameter Surveys• Dynamic Migration: faster/cheaper/bigger machine• Multiple Universe: create job clone to investigate

steered parameter• Automatic Component Discovery and Loading: Needs of job

change• Automatic Convergence Testing• Look Ahead: Spawn off and run coarser resolution to

predict likely future• Spawn Independent/Asynchronous Tasks

– send to cheaper machine, main simulation carries on

• Routine Profiling– best machine/queue, choose resolution parameters based on

queue

• Dynamic Load Balancing and Optimization: inhomogeneous loads, multiple grids, network and cache hierarchies.

Application Scenarios

• Automatically record details about simulations and results.

• Store locations of output data files, along with searchable metadata.

• Ontologies for describing application area: use for meta data, connection to journal publications etc.

• All done in a community accessible way

• SCOOP: SURA Coastal Ocean Observing Program (twiki.sura.org)– NOAA/ONR

• Data coming in from sensors all over Gulf: Realtime Operational Grid

• Feeds in to models on LSU SuperMike and other Grid sites

• Indication of strengthening storm changing direction: Must Act! Multiple simulations, expertise, and data connected together: FL, LA, notification, steering sensors

• Early warning, preparation:– Insurance industry,

petroleum industry, local economy all save millions, lives saved

Hurricanes on the Grid

QuickTime™ and aPhoto decompressor

are needed to see this picture.

LSU, GoMOOS, Texas A&M,

UaH, UNC/MCNC, U. Florida,

U. Miami, VIMS

QuickTime™ and aTIFF (Uncompressed) decompressor

are needed to see this picture.

Ubiquitous Computing and Monitoring System (UCoMS)

• $2.4M DOE/BOR project• ULL, LSU, Southern, Landmark

– Petroleum Engineering, CS, ULL– Deploy sensor networks across Gulf

– Data collected to provide input to simulations, tasks farmed out

– Results collected (transmitted back)

• http://www.ucoms.org

LRZ

Dynamic Grid Computing

Application Tools for Grids

• Application developers need the right kinds of tools and environments to use Grids– APIs, toolkits, debuggers, …

• Focused on the application requirements (MPI, Fortran, legacy, ….)

• Hide the complexities of the Grid• Work in a familiar way (avoid new paradigms)

– Functioning and reliable heterogeneous testbed environments for developing and testing

– Support new and dynamic application scenarios, as generic as possible.

GridLab Project

• EU Funded (5th Framework): Jan 2002– Partners: PSNC, AEI, ZIB, MASARYK, VU, SZTAKI, ISUFI, Cardiff, NTUA, Chicago, ISI, Wisconsin, Sun, Compaq

– GridSphere Portal– Applications

• Numerical Relativity (Cactus)• Gravitational Waves (Triana)• Mobile Users

– Europe-wide Test Bed (Linux, SGI, Hitachi, …)

– Grid Application Toolkit (GAT)• Range of Grid Services

• http://www.gridlab.org

QuickTime™ and aTIFF (LZW) decompressor

are needed to see this picture.

QuickTime™ and aTIFF (Uncompressed) decompressor

are needed to see this picture.

Grid Application Toolkit (GAT)

• Application oriented access to Grid through a standard API– GATFile_Move– GATResource_Find– GATFile_Logical– Independent of Grid infrastructure and available services.

• C, C++, Java, Python, [Perl/Fortran]

• All platforms: Unix, Windows• API driver for GGF SAGA-RG

GAT Motivation• Use of Grids by applications is lagging behind the development of Grid technologies, big jump from prototypes and demonstrations to real production use of Grids needed by applications.

• Problems:– Missing or immature grid services.– Changing environment, new software versions.– Different and evolving interfaces to the “grid”.– “Functioning” Grid is needed to develop applications.

– Interfaces are not simple for scientific application developers.

Monitoring

Resource Management

InformationSecurity

DataManagement

GLOBUS

ApplicationManager

Logging

NotificationMigration

Profiling

SOAP WSDL Corba OGSA Other

Other GridInfrastructure?

Application

“Copy my file from there to there ..”

Example: Remote File Copying

Copy a File: Globus GASS

int RemoteFile::GetFile (char const* source, if (source_url.scheme_type == GLOBUS_URL_SCHEME_GSIFTP || char const* target) { source_url.scheme_type == GLOBUS_URL_SCHEME_FTP ) { globus_url_t source_url; globus_ftp_client_operationattr_init (&source_ftp_attr); globus_io_handle_t dest_io_handle; globus_gass_copy_attr_set_ftp (&source_gass_copy_attr, globus_ftp_client_operationattr_t source_ftp_attr; &source_ftp_attr); globus_result_t result; } globus_gass_transfer_requestattr_t source_gass_attr; else { globus_gass_copy_attr_t source_gass_copy_attr; globus_gass_transfer_requestattr_init (&source_gass_attr, globus_gass_copy_handle_t gass_copy_handle; source_url.scheme); globus_gass_copy_handleattr_t gass_copy_handleattr; globus_gass_copy_attr_set_gass(&source_gass_copy_attr, globus_ftp_client_handleattr_t ftp_handleattr; &source_gass_attr); globus_io_attr_t io_attr; } int output_file = -1; output_file = globus_libc_open ((char*) target, if ( globus_url_parse (source_URL, &source_url) != GLOBUS_SUCCESS ) { O_WRONLY | O_TRUNC | O_CREAT, printf ("can not parse source_URL \"%s\"\n", source_URL); S_IRUSR | S_IWUSR | S_IRGRP | return (-1); S_IWGRP); } if ( output_file == -1 ) { printf ("could not open the file \"%s\"\n", target); if ( source_url.scheme_type != GLOBUS_URL_SCHEME_GSIFTP && return (-1); source_url.scheme_type != GLOBUS_URL_SCHEME_FTP && } source_url.scheme_type != GLOBUS_URL_SCHEME_HTTP && /* convert stdout to be a globus_io_handle */ source_url.scheme_type != GLOBUS_URL_SCHEME_HTTPS ) { if ( globus_io_file_posix_convert (output_file, 0, printf ("can not copy from %s - wrong prot\n", source_URL); &dest_io_handle) return (-1); != GLOBUS_SUCCESS) { } printf ("Error converting the file handle\n"); globus_gass_copy_handleattr_init (&gass_copy_handleattr); return (-1); globus_gass_copy_attr_init (&source_gass_copy_attr); } globus_ftp_client_handleattr_init (&ftp_handleattr); result = globus_gass_copy_register_url_to_handle ( globus_io_fileattr_init (&io_attr); &gass_copy_handle, (char*)source_URL, &source_gass_copy_attr, &dest_io_handle, globus_gass_copy_attr_set_io (&source_gass_copy_attr, &io_attr); my_callback, NULL); &io_attr); if ( result != GLOBUS_SUCCESS ) { globus_gass_copy_handleattr_set_ftp_attr printf ("error: %s\n", globus_object_printable_to_string (&gass_copy_handleattr, (globus_error_get (result))); &ftp_handleattr); return (-1); globus_gass_copy_handle_init (&gass_copy_handle, } &gass_copy_handleattr); globus_url_destroy (&source_url); return (0); }

GAT SolutionAPI layer between applications and the grid infrastructure:

– Higher level than existing grid APIs, hide complexity, abstract grid functionality through application oriented APIs

– Insulate against • Rapid evolution of grid infrastructure• State of Grid deployment

– Choose between different grid infrastructures

– Application developers use and develop for the grid independent of the state of deployment of the grid infrastructure

– Service developers can make their software available to many different applications.

Copy a File: GAT/C++

#include <GAT++.hpp>

GAT::Result RemoteFile::GetFile (GAT::Context context, std::string source_url, std::string target_url) { try { GAT::File file (context, source_url); file.Copy (target_url); } catch (GAT::Exception const &e) { std::cerr << "Some error: " << e.what() << std::endl; return e.Result(); } return GAT_SUCCESS;}

QuickTime™ and aTIFF (LZW) decompressor

are needed to see this picture.

Example: Remote File Copy

GAT:File.Copy

scpcp gsiscp

sftp

gridftp

wgetRFT

GridLabProject Specifi

cGridJITnewest

Application

Grid Application Toolkit

• Standard API and Toolkit for developing portable Grid applications independently of the underlying Grid infrastructure and available services

• Implements the GAT-API– Used by applications

• GAT Adaptors – Connect to capabilities/services– Implements well defined CPI (mirrors GAT-API)– Interchangeable adaptors can be loaded/switched at runtime

• GAT Engine– Provides the function bindings for the GAT-API

GridLab Architecture

GAT API Scope

• Files• Resources• Events• Information exchange• Utility classes (error handling, security,

preferences)

GAT Adaptors

• Default Adaptors: – Basic functionality which will work on a single isolated machine (e.g. cp, fork/exec)

• GridLab Adaptors:– Connect to services developed through the GridLab project (GRMS, Mercury, Delphoi, iGrid)

• Globus Adaptors:– Core Globus functionality: GRAM, MDS, GT-RLS, GridFTP

• Others: DRMAA, Condor, SGE, gsissh, …

GGF SAGA-RG

• Simple API for Grid Applications– Global Grid Forum Research Group– Motivated by GAT-API and other efforts to develop a real common standard• DRMAA, Java-COG, GAT-API, …

– Meeting at LSU Dec04 will provide an initial draft

• A standard API, implemented and supported by vendors, will greatly advance the deployment and adaoption of Grid applications

Simple Examples

Many short examples and “one liners” included in GAT distribution:– Read from a remote file– Read from a logical file– Move a file– Spawn a subtask to a remote machine

– Information about remote/logical file (e.g. file size, permissions)

Example: Spawn a Subtask

GAT::Table sdt; sdt.add ("location", "/bin/date");

GAT::Table hdt; hdt.add ("machine.type", "i686");

GAT::SoftwareDescription sd (sdt);

GAT::HardwareResourceDescription hrd (hdt);

GAT::JobDescription jd (context, sd, hrd);GAT::ResourceBroker rb (context, prefs);

GAT::Job j = rb.submit (jd);

TFM

TFM TFM TFM TFM

Task Farming on the Grid

Task farming infrastructure

implemented in Cactus

GAT used for starting remote TFMs

Designed for the Grid

Tasks can be Anything (MPI, Single proc)

Black Holes on the Grid

• Task farm small Cactus black hole simulations across GGTC Grid

• Parameter survey: black hole corotation parameter

• Results steer a large production black hole simulation

Job Migration

Migration of Legacy Apps

• Using e.g. wrapper scripts the GAT can be used to help Grid-enable “legacy” applications.

Use a GAT wrapper to start

Legacy App

Send a checkpoint

request to the GAT Wrapper

Legacy Application writes

checkpoint file

GAT Wrapper Exits – Kills Legacy App

GAT Wrapper registers

checkpoint file to logical file

GRMS Replicates logical File on

target Host

GRMS starts the GAT Wrapper on

target host

Legacy Application is started within GAT Wrapper

Notification and Information

GridSphere Portal

SMS Server

Mail Server

“The Grid”

Replica Catalog

GridSphere Portal

• Portlet based: JSR 168• Collaboration focal point

• Interact, share data• Start, monitor remote jobs

• Move/browse files• Track and monitor announced jobs

• Application specific portlets– Cactus, UCoMS, SCOOP

QuickTime™ and aTIFF (Uncompressed) decompressorare needed to see this picture.

Portlets• JSR-168 lays the

foundation for a new open-standard for Web portal development frameworks.

• Multiple portlets can be composed in a portal page.

• Portlets define how to construct and deliver Web content as modular components.

• Portlets can be “maximized” or “minimized”.

• Portlets support various modes (View, Edit, Help, Configure)

• Users can subscribe to portlets.

GridSphere Feature List

• Portlets compatible with IBM's WebSphere 4.2.• Support for easy integration of "third-party

portlets"• Higher-level model for building complex portlets.• Customized layouts.• Built-in support for Role Based Access Control

(RBAC) separating users into guests, users, admins and super users.

• Sophisticated portlet service model.• Documentation (DocBook) HTML/PDF guides and

tutorials.• GridSphere core portlets offer base functionality

including login, logout, user and access control management.

• Grid Portlets add general Grid Capabilities.

So what does it look like?

QuickTime™ and aTIFF (LZW) decompressor

are needed to see this picture.

Grid Portlets

• Resource Registry Portlet (Grid/Administration Tab)– Administrators can customize for particular VO

• Resource Browser Portlet (Grid/Resources Tab)– Resource information from MDS/iGrid etc

• Credential Retrieval Portlet (Grid/Credentials Tab)– MyProxy

• Job Submission Portlet (Grid/Jobs Tab)– GRAM, GRMS

• File Browser Portlet (Grid/Files Tab)– GridFTP

• …and many more coming…• Implements generic Grid Portlet API (similar to GAT)

Resource Browsing

Job Submission Portlet

Job History & Output

Links

• GridLab: – http://www.gridlab.org

• Grid Application Toolkit– http://www.gridlab.org/GAT

• GridSphere: – http://www.gridsphere.org

Mardi Gras Conference

Frontiers of Grid Applications and Technologies

Louisiana State University

Baton RougeFebruary 3rd to 5th, 2005

http://www.mardigrasconference.org