90 likes | 181 Views
Grid Education & Outreach. Sanjay Ranka University of Florida S. Masoud Sadjadi Florida International University. http://www.cise.ufl.edu/~lchitnis/TutorialGrid.html. Phase 1: Web Tutorial. Key Features Large scale application focus Evolution from Web Services to Grid Services
E N D
Grid Education & Outreach Sanjay Ranka University of Florida S. Masoud Sadjadi Florida International University
http://www.cise.ufl.edu/~lchitnis/TutorialGrid.html Phase 1: Web Tutorial • Key Features • Large scale application focus • Evolution from Web Services to Grid Services • Practical demos of grid applications using extant infrastructure for Web and Grid services. • Relationship to ongoing standards • 100+ PowerPoint slides • Target Audience • Little or no knowledge of grid computing • Basic programming skills
Phase 1: Representative Activities • Grid ComputingComponent-Based Software Engineering, FIU • 12 hours of lectures on Grid Computing, Web Services, Grid Services, Workflows. • Grid ComputingParallel Computing Course, University of Florida • Grid Computing Concepts were taught in the course on Introduction to parallel Computing. Similarities and differences between parallel computing and grid computing were discussed. • Grid Summer WorkshopSouthPadre Island, TX, June 2006, June 2007 Chitnis and Rodriguez presented a tutorial on building, monitoring and maintaining a grid infrastructure. • Ultralight WorkshopMiami, FL, August 2006 Chitnis presented a tutorial on distributed large scale applications. • FIGS WorkshopMiami, FL, 2008 Chitnis and Rodriguez presented a tutorial on distributing large scale applications. Next workshop in Sao Paulo • Global CyberBridges Miami, FL, 2007-2008 Masoud Sadjadi taught a course on Grid Enablement of Scientific Applications. Attendees at South Padre Island
CHEPREO Phase II (Yr 6-10) We plan to extend the current tutorial and develop educational materials for • storage and networking resources within the grid framework • Service-oriented grids • Meta-scheduling • job flow management • application grid enablement Conduct research in the above areas as a partnership between FIU and UF; and with Brazil
Phase II: Education and Outreach • Storage Resources • use of depots to provide global distributed storage as in L-store [BECK-02, PLANK-01] and • methods for metadata scalability and reliability that are part of systems such as Storage Broker [SRB-03, SRB-98]. • Networks Resources • Salient issues, interfaces and software for providing the high bandwidth interactive and batch traffic • Examples of research networks such as REDDNET, Ultralight and VINCI. • Service Oriented Grids • Common trend and technologies used for realizing service-oriented grids in different research/commercial projects. We will discuss how service-oriented architectures are being used to develop large scale applications through development and composition of Web and Grid services.
Phase II: Education and Outreach • Meta-schedulers • local and meta schedulers • Local scheduler examples: SGE, Condor, Load Leveler • Meta-scheduler examples: GridWay, eNanaos, TDWB • architectural models for meta-scheduling system • centralized, hierarchical, and distributed models • Job Flow Management • how workflows can be used to orchestrate the execution of composite jobs on a grid computing environment. • Examples of workflow languages and systems • Condor DagMan, BPEL+JSDL, Swift, Pegasus, Scufl • Application Grid Enablement • Methodologies, Tools, and Applications • Example Applications • Matrix Multiplication, Weather Research and Forecasting (WRF) and the MOLPRO Computational Chemistry package
Phase II Research: Network Resource Management • Motivation • Emergence of data intensive e-science applications • Transfer of large volumes of data (peta bytes) • Predictable performance • Small high bandwidth networks • Small size (less than 500 nodes) • High bandwidth IP based backbone • Classical QoS provisioning • Low priority best effort delivery for bulk traffic • Centralized network controller • Control plane algorithms • Linear programming driven framework • Multidomain network control 11 9 8 1 10 7 2 3 6 4 5
Phase II Research: Meta Scheduling and Job Flow Management (including Network and Storage Issues) Meteorologist Meteorologist BSC FIU Web-Base Portal Web-Base Portal Job-Flow Manager Job-Flow Manager Peer-to-peer Protocols Meta-Scheduler Meta-Scheduler Resource Policies Resource Policies Loca scheduler Loca scheduler Loca scheduler Loca scheduler Local Resources Local Resources Local Resources Local Resources
Phase II Research • Strong Collaboration between FIU and UF • Co-advising Comp. Science students at the two universities • Work in conjunction with • Global CyberBridges and WHREN-LILA projects at FIU • PLaNets and VINCI projects at Caltech (UF and FIU are participants) • Brazil (Tier-2 Alberto Santoro / Sergio Novaes) • Exchange of graduate students and postgraduate students