Co-ordination & Harmonisation of Advanced e-Infrastructures The CHAIN Project: Technical Achievements Federico Ruggieri, INFN – Project Director NKN Annual Conference 2012 Mumbai, 1 November 2012 Research Infrastructures – Grant Agreement n. 260011
Outline Reasons behind e-Infrastructures and benefits Regional e-Infrastructures The need for a project like CHAIN General information about CHAIN Technical Achievements Conclusions 2
Computing-intensive science Many research challenges require community effort Fundamental properties of matter Genomics Climate change Medical diagnostics Research is increasingly digital, with increasing amounts of data Computation ever more demanding Example: experimental science uses ever more sophisticated sensors Huge amounts of data Serves user communities around the world International collaborations 3 1 st CHAIN Review – Athens, 03 Feb. 2012
The many faces of eInfrastructures The High Speed Communication Network The High Performance Computing for highly parallel applications The Grid for High Throughput Computing and resource sharing The Clouds for elastic resource provisioning Data Infrastructures with several issues such as: Large data volumes, Curation, Access, High Availability, etc. The Human Network: researchers working together sharing motivations, objectives, tools and resources 4
Why ? e- enhance Infrastructures promote international support wide collaboration in collaboration of geographically other fields. scientists distributed communities Grids and networks allow the access of Disparity can be reduced and larger many researchers to participation and contributions to high scientific resources quality research. (laboratories and data) The e-Infrastructures promote the usage of network connectivity and contribute to fight the digital stimulate scientific and technical divide and brain drain. development of countries 5
Regional Grid infrastructures CNGrid EUAsiaGrid NKN & Garuda GISELA SAGrid & SANREN 6
CHAIN: global coverage 7
Project information Grant Agreement for a total EC contribution of 1.1 M € Total cost: about 1.9 M € Start Date: 1st December 2010 - Duration 24 Months Partners: INFN (Italy - Coordinator) 1) CESNET (Czech Rep.) 2) CIEMAT (Spain) 3) GRNET (Greece) 4) IHEP (China) 5) UBUNTUNET (Africa) 6) CLARA (Latin America) 7) PSA (India) 8) ASREN (Med./Middle East/Gulf) Since 1 August 2011 9) 8
Project objectives Define a strategy and a model for external collaboration, in close collaboration with EGI.eu which will enable operational and organisation interfacing of EGI and external eInfrastructures Validate this model, as a proof-of-principle, by supporting the extension and consolidation of worldwide Virtual Research Communities Explore and propose concrete steps forward towards the coordination with other projects and initiatives (e.g. EGI.eu, NKN & Garuda, CNGrid etc.) 9
Project workplan State of the Art Assessment Disseminate (WP5) (WP2) Analyse the different Involve the VRCs Regional Approaches (WP3) (WP2, WP4) Demonstrate the usefulness of interoperation (WP3, WP4) Propose a Road-Map and Make Intermediate solutions Recommendations (WP4, WP3) (WP2, WP3, WP4) 10 1st CHAIN Review – Athens, 03 Feb. 2012
State of the art analysis (WP2) Analysis of existing NGI literature and related questionnaires Creation of the regional and NGI questionnaires Questionnaires being implemented and published online Collection of contact points from all continents Questionnaire is kept open and collection of contact points from all continents is continued Questionnaire data provided through the CHAIN Knowledge Base 11 1st CHAIN Review – Athens, 03 Feb. 2012
Knowledge base (WP2, WP5) www.chain-project/knowledge-base 12
Country view 13
Data analysis (WP4, WP2) Number of sites and number of CPUs Nuber of sites and number of CPU - Asia Pacific 8 7000 CPU cores Sites 7 6000 6 5000 5 4000 4 3000 3 2000 2 Nuber of sites and number of CPU - Latin America Nuber of sites and number of CPU - Mediterranean 1000 1 60 6000 CPU cores 0 0 Sites 8 700 CPU cores Sites 50 Singapore 5000 Indonesia Vietnam New Zealand Taiwan 7 600 6 500 40 4000 5 400 30 3000 4 300 Number of Grid sites / clusters: Number of CPU cores: 3 20 2000 200 2 100 10 1000 1 0 0 0 0 M I A E T r U J a o l g u o n Brazil Ecuador Colombia Argentina Panamá Paraguay Mexico Cuba Venezuela Costa Rica Guatemala Peru g y n n r r i d e o t p i r s e a c i t d a i n a c o A r a b E m i r Number of Grid sites / clusters: Number of CPU cores: a t e s Number of Grid sites / clusters: Number of CPU cores: 14 1st CHAIN Review – Athens, 03 Feb. 2012
WP2 Recommendations 74 Detailed recommendations classified by: Short (1 year), Medium (3 years) and Long term (5 years) High, Medium, Low priority National Grid Initiatives (9): General (5); Regional (4) Interoperations (14): General (1), ROC (3), User Support (1), Monitoring (3), Security (2), Core Services (2); Middleware (2) Interoperability (2): General (1), Input/Output (1) Virtual Research Communities’ perspective: General (2) Regional planning (47): Africa (9), Asia Pacific (6), Central Asia (5), China (7), India (4), Latin America (5), Mediterranean & Arab Countries (11) 15 1 st CHAIN Review – Athens, 03 Feb. 2012
Africa & Arabia Regional Operation Centre http://roc.africa-grid.org 16
VRCs (WP3) Agreements with reference communities signed WeNMR 21/09/2011 WRF4G 19/09/2011 jModelTest 21/02/2012 LSGC 27/03/2012 INDICATE 28/03/2012 DECIDE 13/04/2012 SuperB (on the way) Earth Science (ICTP) (on the way) 17
Worldwide Interoperability Demo 18
Access: the Science Gateway model (WP3, WP4) Administrator Embedded Applications ....... Power User Gateway Basic User App. 1 Science App. 2 App. N Standard-based middleware-independent Grid Engine China Users from Brasil different organisations having different Europe, Africa, Asia India roles and Pacific, Latin America privileges 19
science-gateway.chain-project.eu 20
www.chain-project.eu 21
Clouds and Grids Grid has been conceived as a resource sharing infrastructure Grids provide High Throuput Computing that fits many applications Clouds have a clear business model that allowed companies to provide services on the market Virtualisation and elastic computing are needed by several scientific domains (e.g. interactive and web based applications) Clouds and Grids can cohexist in the Research & Education domain provided that the strenghts of both can be merged 22
R&E Clouds Cloud infrastructure for R&E should be based on Open SW & Standards Financial: Public institutions are frequently receiving projects’ driven funding. Difficult to fund long term contracts with Cloud providers Technical: Resource sharing is an issue if institutions get services from different providers. Building securely across several administrative domains is difficult: Federation of Clouds is still not a reality Long term preservation of data has still many issues to be addressed (e.g. what happens to the data after the project end ?) Requirements of scientists evolve and new technical challenges appear that will push for innovation 23
Conclusions The CHAIN project has gathered the experience and knowledge of regional Grid infrastructures around the world CHAIN has made useful recommendations on several aspects of regional e-Infrastructures and specifically to their sustainability CHAIN has successfully agreed with other projects (EUMEDGRID- Support, GISELA, INDICATE, DECIDE) on the SG approach The first usage of SG in these projects has been very encouraging An Interoperability demo has been demonstrated at the EGI TF 2012 in Prague and is still available (R. Barbera talk in the CHAIN WS of 2 November here at the NKN 2012 Conference in Mumbai) A Road-Map for the interaction between EGI and other regional infrastructures is being finalised New activities will be performed in the CHAIN-REDS project (talk in the CHAIN WS) 24
Co-ordination & Harmonisation of Advanced e-Infrastructures Thank you Research Infrastructures – Grant Agreement n. 260011
Recommend
More recommend
