410 likes | 428 Views
Applications and Cloud Computing at CESGA. Andrés Gómez, PhD CESGA agomez@cesga.es. Show e-science applications on Grid on: Oceanography Health Show how to reuse classrooms Cloud and Science. OBJECTIVES. Portal for Spanish Oceanography Community Requirements:
E N D
Applications and Cloud Computing at CESGA Andrés Gómez, PhD CESGA agomez@cesga.es
Show e-science applications on Grid on: Oceanography Health Show how to reuse classrooms Cloud and Science OBJECTIVES
Portal for Spanish Oceanography Community Requirements: Sharing of resources. Both computing and research instrumentation. Distributed capacity for the storage, management, and exchange of data. Computing capacity for data and image processing. Computing capability. Easy and user-friendly interface. Management of security covering researcher access. Visualization. RETELAB C.Cotelo, et.al. “Retelab: A geospatial grid web laboratory for the oceanographic research community”, http://dx.doi.org/10.1016/j.future.2010.05.018
RETELAB (ARCHITECTURE) C.Cotelo, et.al. “Retelab: A geospatial grid web laboratory for the oceanographic research community”, http://dx.doi.org/10.1016/j.future.2010.05.018
RETELAB (SECURITY) C.Cotelo, et.al. “Retelab: A geospatial grid web laboratory for the oceanographic research community”, http://dx.doi.org/10.1016/j.future.2010.05.018
RETELAB (JOB SUBMISSION) C.Cotelo, et.al. “Retelab: A geospatial grid web laboratory for the oceanographic research community”, http://dx.doi.org/10.1016/j.future.2010.05.018
RETELAB (JOB SUBMISSION) C.Cotelo, et.al. “Retelab: A geospatial grid web laboratory for the oceanographic research community”, http://dx.doi.org/10.1016/j.future.2010.05.018
FORMIGA USING THE CLASS COMPUTER ROOMS TO COMPUTE. ONLY UNIVERSITY OF SANTIAGO HAS MORE THAN 90 CLASS COMPUTER ROOMS WITH MORE THAN 1800 COMPUTERS
FORMIGA (ARCHITECTURE) EACH MACHINE USES AN UNIQUE DIGITAL CERTIFICATE TO ACCESS THE VPN
FORMIGA (i/o) IOZONE Relative Performance (%) Record size (KB) ./iozone -Ra -n 7864320 -g 7864320 -y 4 -q 16384 -b lab.wks -f /mnt/iozone/iozone.tmp
FORMIGA (linpack) LINPACK
DATA CONFIDENTIALITY INTERACTIVITY WORK-FLOW FAST RETURN OF SHORT JOBS HIDE INFRASTRUCTURE DETAILS (PORTALS) HEALTH GRID REQUIREMENTS
Surgical simulation Image processing In silico drug discovery Share patient’s images and data Many more (EELA-2) Source: EELA-2 Source:EGEE Source: Crossgrid project HEALTH GRID APPLICATIONS
60-70% CANCER PATIENTS ESTABLISHED METHODS MANY PROTOCOLS CLINICAL RADIOTHERAPY CLINICAL RADIOTHERAPY
DICOM-CT INCLUDES PATIENT’S DATA IMAGES NEEDS ANONIMYZATION/SECURITY ACCESS FROM GRID TRENCADIS MEDICAL DATA MANAGER (EGEE) DATA ACQUISITION
MEDICAL DATA MANAGER. Insert data J. Montagnat, et.al. ”A Secure Grid Medical Data Manager Interfaced to the gLite Middleware” in Journal of Grid Computing (JGC), 6 (1), pages 45–59, Kluwer, march 2008 Source: EGEE. Johan Montagnat
MEDICAL DATA MANAGER. Get data Source: EGEE. Johan Montagnat
CONFORMAL RADIOTHERAPY (CRT) INTENSITY MODULATED RADIOTHERAPY (IMRT) IMAGE GUIDE RADIOTHERAPY (IGRT) BRACHITHERAPY ETC.
Accelerator Multileaf Collimator CONFORMAL RADIOTHERAPY Tumour is irradiated from several angles Collimator takes the shape of the tumour http://eimrt.cesga.es TUMOUR Organ at risk
INTENSITY MODULATED RADIATION THERAPY (IMRT) Collimator moves during beam-time, modulating intensity Also from different angles TUMOUR Organ at risk
SOFTWARE: TREATMENT PLANNING SYSTEM USE FAST ALGORITHMS RUN LOCALLY: WORKSTATIONS/CLUSTER OUTPUTS (OPTIONAL) DICOM-RTDOSE. CALCULATED DOSE Dose Matrix, Dose Points (2D & 3D), Isodoses, DVH DICOM-RTPLAN: TREATMENT PLAN Fractionation, Tolerance, Patient Setup, Beams, & Sources DOSE CALCULATION
BASED ON BEAMnrc and DOSXYZ MONTE CARLO CALCUTES DOSE DISTRIBUTIONS FOR AN EXISTING TREATMENT ADDS TOOLS FOR COMPARING REFERENCE AND CALCULATED DOSES (3D GAMMA MAPS) PLAN VERIFICATION PLAN VERIFICATION. EIMRT SERVICE D.M. González-Castaño, et. al. “eIMRT: a web platform for the verification and optimization of radiation treatment plans”, Journal of Applied Clinical Medical Physics, Vol 10, No 3 (2009)
Results Results • Commisioning • Verification < 5 hours • Optimization < minutes E-IMRT PROPOSAL CTs Treatment BEFORE TPS WITH E-IMRT
PERSONAL DATA REMOVED FROM INPUT FILES BEFORE UPLOAD E-IMRT ARCHITECTURE SERVER CLIENT GRID + CLUSTER Service Oriented Architecture Based on GRID technologies
GRIDWAY E-IMRT ARCHITECTURE SERVER SIDE DEMO CLIENT DRMAA SLA SOA Architecture Based on GRID technologies
SLA Negotiator client SLA NEGOTIATION OVERVIEW GRID TREATMENT SERVICES GRIDWAY DRMAA SLA SLA Negotiator server EXTERNAL RESOURCES PROVIDER
SLA Negotiator client SLA COMPONENTS Provider List Broker GW-SLA SLA Negotiator server Pre SLA GW Internal Struct SLA Evaluation Resources provider DB Services Plugin GW-SLA GRIDWAY M.G. Bugeiro, J.C. Mouriño, A. Gómez, C. Váquez, E. Huedo, I.M. Llorente, D.A. Rodríguez–Silva “Integration of SLAs with GridWay in BEinEIMRT project” Conference 3rd Iberian Grid Infrastructure Conference (IBERGRID 2009) Valencia (Spain), May, 2009 ISBN 978849745406-3
IMRT TREATMENT VERIFICATION • Phase 1: Accelerator simulation. • Phase 2: Accelerator treatment head simulation (GRID) • Phase 3: Patient simulation. • Phase 4: Dose delivered to the patient(GRID) • Phase 5: Dose collection and end of process. Radiotherapist manually compares TPS and e-IMRT Monte Carlo doses Using different maps
IMRT TREATMENT VERIFICATION • BUT: • 6200 Radiotherapy institutions in the world. • 618 in Europe • 10% market share = 620 hospitals • 1/3 in Europe = 200 Hospitals • 200 Treatments to verify x 50 CPUs/treatment = 10.000 CPUs allocated = small Supercomputer Center. • Cloud solution? • Avoid interferences between treatments • One virtual cluster/treatment • Secure storage
Cloud Computing Grid Computing Grid Computing vs Cloud Computing Source:Trends.google.com
“Clouds are a large pool of easily usable and accessible virtualized resources (such as hardware, development platforms and/or services). These resources can be dynamically reconfigured to adjust to a variable load (scale), allowing also for an optimum resource utilization. This pool of resources is typically exploited by a pay-per-use model in which guarantees are offered by the Infrastructure Provider by means of customized SLAs.” Luis M. Vaquero, et.al.: “A Break in the Clouds: Towards a Cloud Definition“ CLOUD COMPUTING DEFINITION
CLOUD COMPUTING PERFORMANCE http://www.paessler.com/blog/2009/04/06/prtg-7/monitoring-cloud-computing-performance-with-prtg-cpu-disk-memory-speed-comparison-of-amazon-ec2-instance-types/
CLOUD COMPUTING PERFORMANCE Keith R. Jackson, Karl J. Runge, Rollin C. Thomas “Seeking Supernovae in the Clouds: A Performance Study”
Normalized Usage of Business-oriented Architectures Proyecto parcialmente subvencionado por el subprograma Avanza I+D de la Acción Estratégica de Telecomunicaciones y Sociedad de la Información del Ministerio de Industria, Turismo y Comercio de España. Número de proyecto: TSI-020301-2009-30 43 43
Development of federated Cloud IaaS (Infrastructure as a Service) for the deployment of services permiting the dynamic scalability based on performance and business objectives. 44
2. Storage virtualizacion Front-end virtualization 1.Cluster virtualization E-IMRT ON CLOUD SERVER CLIENT Service Oriented Architecture NUBA TSI-020301-2009-30 FUNDED BY
E-IMRT ON CLOUD CE1 CPU & 1GB Master Node 1CPU & 1GB Scratch 2GB Grid Engine Home (NFS) 12GB Phases 1, 3 and 5 Scratch 2GB Phase 2 and 4 DOSXYZnrc
THANK YOU Any questions?