The EU-IndiaGrid Project Joining European and Indian grids for e-science

1. EELA WP2 E-infrastructure Workshop Rio de Janeiro, 20-23 August 2007 The EU-IndiaGrid ProjectJoining European and Indian grids for e-science Marco Verlato (INFN-Padova)

2. Outline • Project overview • Project status @ month 10 • What next

3. An EGEE related project

4. Eu-Indiagrid Project Goals • To support the interconnection and interoperability of the prominent European Grid infrastructure (EGEE) with the Indian Grid infrastructure for the benefit of eScience applications; • To identify and aggregate research, scientific and industrial communities which may benefit from the use of Grid technology resulting in an eScience Network Community • To promote the use of advanced Grid technologies within the created Network Community relying on pilot applications in Biology, High Energy Physics, Material Science and Earth and Atmospheric Sciences and specific outreach and dissemination activities • To disseminate European EGEE Grid technology achievements in India and leverage on Indian Grid experiences and skills.

5. PARTICIPANTS EUROPE • INFN (project coordinator) • Metaware SpA • Italian Academic and Research Network (GARR) • Cambridge UniversityINTERNATIONAL • Abdu Salam International Centre for Theoretical PhysicsINDIA • Indian Education and Research Network (ERNET) • University of Pune • SAHA Institute of Nuclear Physics, Kolkata • Centre for Development of Advanced Computing (C-DAC) • Bhabha Atomic Research Centre, Mumbai • TATA Institute for Fundamental Research (Mumbai) National Centre for Biological Sciences (Bangalore)

6. Project Effort • BUDGET: • 1208 k-EUR total fund • 1015.9 k-EUR from European Commission • Person months • 353.3 PM total • 226.4 PM funded from European Commission • Project start: October 2006 • Duration: 24 months

7. WP3 WP3 WORKPACKAGES WP5 Garuda-India EGEE WP4 WP2 WP6

8. Project Status @ month 10 • Two Project Workshops so far • Mumbai, December 1-4 2006, jointly organized with the WLCG Tier2 in Asia Workshop • Focused on High Energy Phyisics applications • Over 100 registered participants • Bangalore, April 16-18 2007 • Focused on new communities and application domains • Over 40 invited participants, representatives from Research, Academia and Industry • Boost for all project activities • Building an eScience Network Community • Network Planning & Support • Pilot grid infrastructure operational support • Applications • Dissemination

9. WP2-Building an eScience Network Community • Contacts & collaborations already established at the Project Workshop and at the EU-IndiaGrid session during the Belief Conference in December, and then at G8-Unesco meeting of May 2007 in Trieste • Identify targets • 1400 contacts, active grid users from India & Europe • beyond the four application areas directly addressed • Focus on building the Indian Community of contacts Over 400 members from about 250 Institutions registered through the EU-IndiaGrid portal at 30 June 2007 • All Registered members inserted in the EU-IndiaGrid DB and required to fill a questionnaire • 68% from India, 22% from Europe, 1% from Asia, 9% from other countries

10. WP3-Network Planning & Support • ERNET: Education and Research Network • An Autonomous Organization under administrative control of Department of Information Technology, Govt. of India • 15 Point of presence (POPs) at premier E&R institutions in India • 50 E&R institutes interconnected in 17 cities of India • Leased line, VSAT • MPLS enabled backbone • Satellite Hub in C-band at Bangalore • 170 Mbps Internet gateway bandwidth • maintain Internet domain registration for “ac.in”, “res.in” and “edu.in”

11. WP3-Network Planning & Support • A layer 3 MPLS VPN outsourced to a commercial provider • 45 Institutes in 17 cities of the country • Access speed of 10 Mbps for 23 sites and 100 Mbps for 22 sites • Backbone bandwidth of 2.43 Gbps • Managed network • Strict Service Level Agreement parameters

12. WP3-Network Planning & Support • Analysis of the network status EU-India and within India to ensure adequate access to the intercontinental network • GEANT-ERNET Milan-Mumbai 45 Mb/s link opened in October 2006 • Issues &actions • GARUDA-Grid sites not connected to the GEANT-ERNET link @ project month 3 • Ensured access of GARUDA-Grid to GEANT interconnecting two routers at ERNET PoP in Mumbai hosted by C-DAC • Insufficient bandwidth to access some sites • Upgraded the link C-DAC Mumbai to GARUDA from 10 Mb/s to 100 Mb/s • Need for an upgrade of the intercontinental link once the applications will exploit the grid infrastructure • Support actions targeted at upgrading the intercontinental link

13. EU-IndiaGrid network Garuda Cloud Tier2 ALICE VECC KOLKATA 10100 Mbps 34 Mbps 45 Mbps IPLC GEANT POP AT MILAN ERNET POP AT C-DAC MUMBAI 10 Mbps VPN Cloud connecting Punjab University, Aligarh University, University of Rajasthan, Jammu University, Institute of Physics 34 Mbps 34 Mbps ERNET Cloud Tier3 CMS BARC MUMBAI Tier2 CMS TIFR MUMBAI

14. WP4-Pilot grid infrastructure operational support • Key points • Certification Authorithy • Creation of a pilot testbed • Interoperability

15. Certification Authority • Cooperation with Academia Sinica (Regional Operation Centre for Asia) allowed: • To established Registration Authorities at each site in order to ensure immediate grid access worldwide to Indian users • Set the necessary steps for an Internationally recognized Indian Certification Authority(Responsibility taken by C-DAC)

16. Pilot Testbed

17. Available Services Since December 2006: • 1 VOMS server @ INFN-CNAF (Bologna)https://voms2.cnaf.infn.it:8443/voms/euindia • 1 LFC server @ INFN-CNAF (Bologna)lfcserver.cnaf.infn.it • 1 TOP LEVEL BDII @ INFN-PADOVAeu-india-03.pd.infn.it • 1 WMS-LB (RB) @ INFN-PADOVAeu-india-02.pd.infn.it • 1 GridICE server @ INFN-PADOVAhttp://euindia-gridice.pd.infn.it:50080/gridice/ • 1 Gstat@ AP-ROC, Taiwanhttp://gstat.gridops.org/gstat/euindia

18. 4 Mbps Link 4 Mbps Link CAT: archival storage VECC: real-time Data collection BARC: Computing with shared controls IGCAR: wide-area Data dissemination DAE Grid Interoperability • India has two main grid infrastructures: • The GARUDA National Grid Initiative • The Department of Atomic Energy (DAE) Grid 100 Mbps and 10 Mbps links GARUDA Grid

19. GARUDA/Sigmacomponents

20. Interoperability • Issues and actions deeply discussed at the Project Workshops: • Pilot testbed based on gLite • Gridway proposed to address interoperability between GARUDA and gLite (cooperation between C-DAC and Madras Institute of Technology) • But a wider vision of interoperability based on OGF standards is needed, profiting of the outcome of EU projects like OMII-Europe (JSDL, BES, VOMS/SAML, GLUE, etc.) • EU-ChinaGrid faced similar issues  joint workshops at EGEE’07 and IEEE eScience2007 conferences

21. Example: JSDL/BES scenario in OMII-EUROPE

22. WP5-Applications • Good progress for all the four application areas directly addressed by the project • Boost from the Mumbai and Bangalore Workshops and follow-up meetings, especially for High Energy and Condensed Matter Physics, Earth and Atmospheric Sciences applications

23. Applications: Atmospheric sciences • Collaboration between: • Physic of Weather and Climate group at Abdus Salam International Centre for Theoretical Physics (ICTP) in Trieste, Italy • Computational Atmospheric Science group at Centre for Development of Advanced Computing (C-DAC) in Pune, India • The first challenges: • Develop a Regional coupled atmosphere-ocean model system for “Prediction of the Indian summer Monsoon” • Coupled meteorology-air quality modeling system for the study of pollutant dispersion in urban areas • Simulation Tools • Several atmospheric/climate models • MM5/RegCM3/ROMS/WRF etc... • First step: port them on the EU-IndiaGrid

24. The RegCM3 application • RegCM3: Regional Climate Model Version 3 • CPU intensive • tightly coupled parallel code • I/O intensive code • EX: The model generates about 9GB of data for one month of model integration for the domain covering the monsoon region • the RegCM3 model is being integrated for a 15 years period to study the intraseasonal and interannual variation of the Indian summer monsoon • It has been adapted and ported on both the gLite based and Garuda grids

25. Applications: Condensed Matter • Many different research groups within University of Pune and ICTP • Different applications/porting approaches: from trivial parameter-sweep computational tasks to tightly coupled MPI parallel programs • Real Space Density Function Code for Quantum dots (University of Pune) • Equilibrium geometries of atomic clusters using VASP code (University of Pune) • Quantum ESPRESSO suite for electronic-structure calculations and materials modeling at the nanoscale (ICTP) • Biased exchange metadynamics alogorithm based on Gromacs MD package (ICTP)

26. Electronic structures of QDs Many iterations with many initial guesses, each with different spin and impurity parameters  thousands of independent jobs

27. Quantum ESPRESSO suite porting • Tightly coupled MPI parallel program • Doesn’t fit with MPICH over Ethernet implementation typical of gLite based resources (latency and bandwidth limitations) • New approach: use most recent multicore WNs to run small parallel jobs exploiting shared memory for communication • Python utility developed to reserve a full WN on a CE, more details at: http://partners.euindiagrid.eu/deliverables/D5.2.html

28. Applications: Biology • Focused on porting to grid the Multiscale Object-Oriented Simulation Environment (MOOSE) • Work lead by NCBS team in Bangalore • Compute intensive application • Some models partitionable in loosely coupled sub-models which can run independently

29. Applications: High Energy Physics • HEP LHC applications are already grid-enabled • EU-IndiaGrid is supporting the establishment and the operations of two WLCG Tier-2 sites in India: • TIFR/Mumbai involved in CMS • SAHA/VECC Kolkata involved in ALICE • First file transfer from CERN to Mumbai and Kolkata exploiting the 45 Mbps link initiated the 12 February 2007 (see http://www.isgtw.org/?pid=1000300)

30. Applications: High Energy Physics ALICE Physics Data Challenge at SAHA/VECC Kolkata Tier-2

31. WP6-Dissemination • Promotion material • Several hundreds contacts reached through participation to Conferences (G8, BELIEF-EELA, EGEE, OGF, etc.) • Thousands contacts reached through press releases • Website • More than 400 contacts from about 250 Institutions registered in the EU-IndiaGrid Community through the portal • Training events • 1° Training Event during the EU-IndiaGrid – WLCG Mumbai Workshop, December 2006 • 2° Training Event during the EU-IndiaGrid Interactive Workshop in Bangalore, April 2007 • 3° Training Event for Site Administrators at Kolkata, June 2007

32. Next Steps • Towards strengthening and enlarging the collaboration • Towards Indian National Certification Authority recognized internationally • Toward an EU-IndiaGrid testbed involving all sites • Towards an operational service for Applications • Towards GARUDA-gLite interoperability

33. Summary and plans for EU-FP7 Less than one year from its start-up the project has: • given access to Indian Researchers to the Intercontinental Grid Infrastructure (on temporary basis through a cooperation with Academia Sinica Taiwan) and taken the necessary steps for an internationally recognised Indian Certification Authority. The selected Institution is C-DAC, spearhead of GARUDA Indian National Grid Initiative • promoted and supported cooperation in grid technology between Europe and India • identified and aggregated a community of about 400 people from over 250 Institutions in Research, Academy and Industry • promoted the use of grid technology through its supported four main application areas and identified new applications interested in the use of the grid infrastructure • set-up a test-bed for the benefit of EU-IndiaGrid applications

34. Summary and plans for EU-FP7 On this basis, an new EU-FP7I3 project EU-IndiaGrid2 is proposed, with the following objectives: • development of the appropriate tools to ensure a full interoperability between the European and Indian grid infrastructures • cooperation on middleware developments of common interest for EGEE and the GARUDA National Grid Initiative • support and operation of National Indian Certification Authority internationally recognised so to give access for Indian researchers to the intercontinental grid infrastructure • support an Indian Regional Operation Centre through a cooperation with the GARUDA National Grid Initiative in order to ensure a production service of the grid infrastructure to the applications (including pilot cases of business applications) • continue the dissemination and training activity supporting the existing applications and identifying new application areas where European and Indian collaborations may benefit of grid technologies

35. Future Events • September, 24-28, Pune - Project Workshop focused to Condensed Matter Physics, Earth and Atmospheric Sciences applications and Training Event • December, 13-14, Bangalore–Project Conferencein conjunction with the 3rd IEEE International Conference on eScience and Grid Computing

36. Thank you