A Service-oriented Grid Infrastructure for Biomedical Data and ... - Nettab

5 downloads 1030 Views 621KB Size Report
NETTAB'06. Santa Margherita di Pula, July 10-13, 2006. A Service-oriented Grid Infrastructure for. Biomedical Data and Compute Services. Siegfried Benkner.
A Service-oriented Grid Infrastructure for Biomedical Data and Compute Services Siegfried Benkner Institute of Scientific Computing University of Vienna AUSTRIA Co-Authors: T. Arbona, J. Fingberg, G. Engelbrecht, M. Hofmann, K. Kumpf, G. Lonsdale, A. Wöhrer

S.Benkner, University of Vienna

NETTAB’06

Santa Margherita di Pula, July 10-13, 2006

Outline ‰ The EU Integrated Project @neurIST ‰ @neurIST Grid Architecture and Middleware ‰ Compute Services ‰ Data Services ‰ Conclusions and Future Work

S.Benkner, University of Vienna

NETTAB’06

Santa Margherita di Pula, July 10-13, 2006

EU Project @neurIST (2006-2009) Integrated Biomedical Informatics for the Management of Cerebral Aneurysms 33 Partners, ~17,5 MEuro;

http://www.aneurist.org

† Development of a generic IT infrastructure for the management and processing of heterogeneous data for diagnosis and treatment of cerebral aneurysms. † On-demand simulation and data-integration services handling multi-scale, multi-modal information at distributed sites

Source: Peter Hunter Human Physiome Project S.Benkner, University of Vienna

NETTAB’06

Santa Margherita di Pula, July 10-13, 2006

@neurIST Integrative Application Suites

@neurIST Systems

@neuLink

Support Tools

@neuEndo

Support computational design processes towards a next generation of smart flowcorrecting implants to treat ruptured aneurysms and reduce current treatment costs, side effects and recurrence. Support the knowledge discovery for linking genetics to disease, vasospasm and blood clotting after cerebral hemorrhage WSS

@neuFuse

Support the integration of modeling, simulation and visualization of multimodal data

Mod &W

CF Peak

3DRA

PC-MR vs US Flow rates magnitude

phase

CFD

@neuCompute/Info

S.Benkner, University of Vienna

Support integration of data and computing resources.

NETTAB’06

Enabling IT

@neuRisk

Improve decision making processes in the management of unruptured aneurysms by providing a score that integrates all the available information for identifying at-risk patients and reducing current over treatment

Santa Margherita di Pula, July 10-13, 2006

@neurIST Integrative Application Suites

Patient Imaging data

Gen data

Personal data Simulation services

@neuFuse

Epidemiology knowledge

S.Benkner, University of Vienna

@neuLink

Genetic risk databases

@neuRisk

Pharma treatment NETTAB’06

@neuEndo Santa Margherita di Pula, July 10-13, 2006

@neurIST Service-Oriented Grid Architecture Integrative Application Suites @neuRisk

@neuLink

@neuEndo

@neuFuse

Integrative rupture risk assesment

Linking genetics to disease

Virtual endovascular treatment planning

Multimodal data processing & fusion

Computing and Infostructure @neuCompute

@neuInfo

Computing and data transport services

Distributed and heterogeneous data access

Job & data handling interface Compute services

Data service interface

Data staging services

Data services

Data Mediation

OGSA−DAI

GDMS

Core Grid Middleware: GEMSS and InnerGrid Information and Compute Resources Public databases Hospital Information Genetic: EBI,NCBI Literature: Medline, systems etc.

S.Benkner, University of Vienna

HUG, USFD, etc.

Private databases COTS stents, coils, etc

NETTAB’06

Modelling and Simulation Applications Santa Margherita di Pula, July 10-13, 2006

GEMSS - Architecture & Middleware ‰ Service oriented architecture

CA

Client - Based on standard Web Services technology. - WSDL, SOAP, WSRF - End-to-End Security (own WS-Security implementation)

Service Registry

‰ Service-Provision Framework

- Virtualization of HPC applications/data sources as services - Application-level QoS guarantees (response time, price, location, …)

‰ Grid Programming Framework

- Construction of Client-side Grid applications using High-Level APIs

‰ Service access model Business Phase

QoS Negotiation Phase

AAA Security Pricing Model

Estimate job capacity Negotiate QoS Exchange Contracts

S.Benkner, University of Vienna

NETTAB’06

Job Handling Phase Upload input data Start job Monitor job Download results

Santa Margherita di Pula, July 10-13, 2006

Uniform Grid Services Applications and data virtualized as Grid/Web Services with uniform interface (WSDL) and protocol/transport (SOAP/HTTP).

Compute services negotiate upload Grid client

start download

WSDL negQoS() upload() start() download() …

App

Data services upload Grid client

start download

S.Benkner, University of Vienna

NETTAB’06

WSDL upload() start() download() …

DB DB DB

Santa Margherita di Pula, July 10-13, 2006

Service Provision Infrastructure ‰ Hide complexity of the Grid from the Grid service providers/developers ‰ End-to-end security; Firewall-friendly.

Compute Service

Scheduler

Authorization

Web Service Security

Apache Web Server (with OpenSSL)

Firewall

‰ Deploytool automates provision of applications/ data sources as Grid services

Application

MPI

Data Service

Files

DB x509v3 Certificate

S.Benkner, University of Vienna

Service Container (Tomcat,Axis) NETTAB’06

DB Santa Margherita di Pula, July 10-13, 2006

HPC Application

Grid Application Service

(1) Install HPC application on a Grid host executable + libraries, etc (2) Specify Application Descriptor (Deploytool) I/O files, job script, status script, resource constraints (3) Specify QoS Attributes, QoS Models, and Descriptors - e.g.:Performance model IN: Request Descriptor (input meta data, # processors) OUT: Performance Descriptor (run time, memory, etc.) - Price model (5) Deploy application service (Deploytool) S.Benkner, University of Vienna

NETTAB’06

Santa Margherita di Pula, July 10-13, 2006

On Demand Supercomputing † Client driven QoS negotiation with potential service providers Client supplies: QoS requirements (e.g. response time), Input meta data

† Signed Web Service Level Agreement (WSLA) exchanged with winner † Advance reservation of resources Start: 2:00 End: 2:30 Price: 8€

Client Application Request N=10000 QoS Start: 2:00 End: 2:20 Price: 10€

Ask for WSLA offers Offer WSLA

QoS Negotiator

Accept/ Reject offers Start: 2:00 End: 2:18 Price: 11€ WSLA offered

S.Benkner, University of Vienna

NETTAB’06

Application Service QoS Manager



App

Resources

Service Provider N

Application Service QoS Manager

Scheduler

WSLA offered

Scheduler

Service Provider 1

App

Resources

Santa Margherita di Pula, July 10-13, 2006

Data Services • Same interface and access/security mechanisms as compute services • Utilize OGSA-DAI to hide heterogeneity of data sources • Can be configured for data mediation by utilizing GDMS • Will be extended with OGSA-DQP to optimize distributed queries Data Service DQP GDMS OGSA/DAI

Relational DB S.Benkner, University of Vienna

XML DB NETTAB’06

CSV Files Santa Margherita di Pula, July 10-13, 2006

start

upload

OGSA-DAI Perform Document

download

Data Services – Mediation (GDMS) OGSA-DAI Response Document

• Transparent access to multiple data sources

Data Service

• GEMSS security & transport

GDMS Mapping Schema

• OGSA/DAI access mechanisms

GDMS

w.r.t. virtual DB (global schema)

GDMS Transform. Functions

OGSA/DAI

• read-only access !!

Virtual DB S.Benkner, University of Vienna

Relational DB NETTAB’06

XML DB Santa Margherita di Pula, July 10-13, 2006

Conclusions @neurIST Project ‰ Development of a generic Grid infrastructure for the management and processing of heterogeneous data for diagnosis and treatment. ‰ SLA-based on-demand simulation and data-integration services handling multi-scale, multi-modal information at distributed sites.

@neurIST Grid Environment ‰ Compute and Data services with uniform interface based on WS-Standards ‰ leverages GEMSS, InnerGrid and OGSA-DAI/DQP developments

Challenges ‰ Semantic Grid Infrastructure, Ontologies ‰ Security, legal issues, provenance wrt. to patient data S.Benkner, University of Vienna

NETTAB’06

Santa Margherita di Pula, July 10-13, 2006

@neurIST Consortium

www.aneurist.org

ANSYS Europe Ltd., UK

NEC Europe Ltd., DE

Advanced Simulation and Design GmbH, DE

Philips Medical Systems, NL

Cancer Research UK

Supercomputing Solutions S.R.L., IT

Ecole Polytechnique Federale de Lausanne, CH

The Thrombosis Research Institute, UK

Erasmus University Medical Centre, Rotterdam, NL

U. Clinic Freiburg, DE

FIMIM, Foundation IMIM, ES

U. Geneva (and U. Hospital), CH

Fraunhofer Gesellschaft (SCAI), DE

U. Luton, UK

Grid Systems S.A., ES

U. Medical Centre, Utrecht, NL

IDAC Ltd., IE

U. Oxford, UK

IMIM (Inst. Municipal d’Assistencia Sanitaria), ES

U. Pompeu Fabra, ES (Coordinator)

Infermed Ltd., UK

U. Sheffield, UK

INSERM,FR

U. Vienna, Austria

Hospital “Clinic I Provincial de” Barcelona, ES

William Cook Europe APS, Denmark

KTH (Kungliga Technische Hoegskolan), SE

Tohoku U., JP

Medical U. Pecs, HU

George Mason, US

Neuroangiografia Terepeutica S.L, ES

Mayo Clinic, US Centre for Biomolecular Discovery, U. Wellington, NZ

S.Benkner, University of Vienna

NETTAB’06

Santa Margherita di Pula, July 10-13, 2006

Suggest Documents