1. EGEE-III INFSO-RI-222667 Enabling Grids for E-sciencE www.eu-egee.org EGEE and gLite are registered trademarks Steven Newhouse OGF 25, Catania Defining e-Infrastructure for the masses

2. Enabling Grids for E-sciencE EGEE-III INFSO-RI-222667 The Pioneering Life Cycle Defining e-Infrastructure for the masses - OGF 25 InnovateStandardiseExploit 2

3. Enabling Grids for E-sciencE EGEE-III INFSO-RI-222667... and with e-Infrastructure? Innovate: The last century –Software: Globus, Legion, Condor,... –Experiments: I-Way, meta-computing,... Standardise: The naughty noughties –Experience: European Data Grid, OSG, NAREGI, PRAGMA –Bodies: Global (  Open) Grid Forum & Enterprise Grid Alliance –Specifications: Information, Compute,...  The specification ‘flowers’ that bloomed while others did not... Exploit: The coming terrific teens –Reliability: Having something that works –Availability: Using something that’s there –Dependability: A foundation for multiple activities Defining e-Infrastructure for the masses - OGF 25 3

4. Enabling Grids for E-sciencE EGEE-III INFSO-RI-222667 International e-Infrastructure Defining e-Infrastructure for the masses - OGF 25 Enabling Grids for E-sciencE is the largest multi-disciplinary grid infrastructure in the world. It brings together more than 140 institutions. At present, it consists of approximately 300 sites in 50 countries and gives its 10,000 users access to 80,000 CPU cores around-the-clock. The grid infrastructure currently processes up to 300,000 jobs per day from scientific domains ranging from biomedicine to fusion science. 4

5. Enabling Grids for E-sciencE EGEE-III INFSO-RI-222667 What have we learned? Defining e-Infrastructure for the masses - OGF 25 Community e-Resource Life used to be simple.... Community e-Resource Then it got more complicated.... Community e-Resource Then it got really complicated.... shared use of different types of e-resources Community e-Resource 5

6. Enabling Grids for E-sciencE EGEE-III INFSO-RI-222667 Admin Domain B Admin Domain A We need some e-infrastructure! Defining e-Infrastructure for the masses - OGF 25 e-Infrastructure Community Storage Resource HTC & HPC Resource Instrument Resource Data Resource Other Resources Accounting Monitoring Job Management AuthenticationAuthorization File Management Information 6

7. Enabling Grids for E-sciencE EGEE-III INFSO-RI-222667 So why do we need standards? Standards are a fundamental part of our lives –Standard shoe and clothes sizes –Mobile phone networks Why should e-infrastructure be any different? To enable an open world for flexible collaboration –No central control of the infrastructure –Portability of applications and software –Encompass different perspectives on the ‘best’ solutions Standards provide a route to independent collaboration –Plugin and swap out components over time –A structured open mechanism to evolving common interfaces Defining e-Infrastructure for the masses - OGF 25 7

8. Enabling Grids for E-sciencE EGEE-III INFSO-RI-222667 GLUE 2.0 GLUE: Grid Laboratory Uniform Environment –Started in April 2002 –Join activity between EU-DataTAG, US-iVDGL and EDG  Focused on interoperability between US and EU HEP projects –Aimed to provide common schema to facilitate interoperations –v1.0 released Nov 2002 Work within the Open Grid Forum –October 2006: First discussion of moving work into OGF –January 2007: Working group created and first meeting held –June 2008: Specification entered public comment –March 2009: GLUE 2.0 approved as proposed recommendation Defining e-Infrastructure for the masses - OGF 25 8

9. Enabling Grids for E-sciencE EGEE-III INFSO-RI-222667 Creating GLUE 2.0 Support from many Organisations & Projects –EGEE, ARC, UNICORE, Platform, OSG, APAC, NGS, NAREGI, OMII-Europe, OGF-Europe GLUE co-chairs –Balazs Konya (Lund University) –Sergio Andreozzi (INFN) –Laurence Field (CERN) Core GLUE Attendees: –Stephen Burke (RAL), Maarten Litmaath (CERN) –Paul Millar (DESY), JP Navarro (ANL) Defining e-Infrastructure for the masses - OGF 25 Sergio Andreozzi GLUE 2.0: Getting To Public Comment 347 days to produce the initial specification 45 phone conferences (~ 3 days) Core team of 5 people (~ 2 months FTE) 40 versions of the document (~ weekly updates) 9

10. Enabling Grids for E-sciencE EGEE-III INFSO-RI-222667 Key Concepts in GLUE 2.0 Resources are provided and managed by Administrative Domains. –Universities, research institutes and private companies are all examples of Administrative Domains that provide Resources. User Domains represent users who which to utilize a Resource. –Individuals, projects, communities and Virtual Organizations are examples of User Domains. A User Domain negotiates with an Administration Domain to gain access to the Resource –This may result in a Service Level Agreement. –Members of the User Domain are granted access and usage of a Share of the Resource. –May be implemented through a Management entity which manages the Resource. Defining e-Infrastructure for the masses - OGF 25 10

11. Enabling Grids for E-sciencE EGEE-III INFSO-RI-222667 GLUE 2.0 Key Concepts User Domain Admin Domain Resource Manager ShareEnd Point Activity Access Policy Mapping Policy Negotiates Share with Provides Manages Runs Defined on Contacts Maps User to Has Service Defining e-Infrastructure for the masses - OGF 25 11

12. Enabling Grids for E-sciencE EGEE-III INFSO-RI-222667 Using GLUE Defines an extensible abstract usage model Concrete usage models in the specification: –Compute resources –Storage resources User Domains access resources via a Service End Point. –This End Point requires an Access Policy and Mapping Policy to ensure the User Domain is mapped to the correct Share. The User Domain is then able to run an Activity on the Resource which is accounted to correct Share. Defining e-Infrastructure for the masses - OGF 25 12

13. Enabling Grids for E-sciencE EGEE-III INFSO-RI-222667 Standards within EGEE Increasing focus on interoperability –Driven implicitly by the user community –Their need to use resources within different e-Infrastructures  EGEE with Open Science Grid and the Nordic Data Grid Facility Capturing our intellectual knowledge –A lot of time and effort has been invested in middleware –Standards are one approach to expressing know how –Provides a return on the EU investment –Knowledge transfer to EU business community A middleware exit strategy –Define its functional behaviour – the standard interface –Define its non-functional behaviour – performance characteristics Defining e-Infrastructure for the masses - OGF 25 13

14. Enabling Grids for E-sciencE EGEE-III INFSO-RI-222667 Standards within EGI Opening up middleware provision to other providers –Work with software providers in the EU, worldwide & industry NGIs have subsidiarity in the software they deploy –EGI provides a solution that works across common environments  NGIs are free to substitute alternatives  Comply with same interface & minimal performance capabilities Emerging need to independently assess software –One of the original goals from the OMII-Europe project –Is this a role for OGF-Europe to move into in? Defining e-Infrastructure for the masses - OGF 25 14

15. Enabling Grids for E-sciencE EGEE-III INFSO-RI-222667 Extensible & Open e-Infrastructure Monitoring Monitoring of Site Service Level Agreements User Community Users National Infrastructure Services Site Services User Community Users Community Platform Services Jobs Registry Files Security Jobs Registry Security National Infrastructure Services Site Services National Infrastructure Services Site Services Files Software Services Dashboards Workflows Defining e-Infrastructure for the masses - OGF 25 15

16. Enabling Grids for E-sciencE EGEE-III INFSO-RI-222667 NGI Site Services Where do we need standards? Defining e-Infrastructure for the masses - OGF 25 User Community (Desktop Applications, Portals, Project Services) Compute Storage Information Compute Storage Information Community Services Compute Storage Information Compute Storage Information Reporting Services Files & Jobs Accounting Authorization 16

17. Enabling Grids for E-sciencE EGEE-III INFSO-RI-222667 Standards to enable Federation Authorization Services: –X.509 Certificates, SAML, XACML Information & Reporting Services: –Grid Laboratory Uniform Environment (GLUE) –Usage Records (Describing Accounting Activity) –Resource Usage Service (Exporting Accounting Records)* Compute Services: –Basic Execution Services (Job Submission Interface) –Job Submission Description Language Storage Services –GridFTP (Data Transport) & Storage Resource Management –Database Access Integration Services (DAIS-WG)* –Data Movement Interface (File Transfer)* * Specifications potentially relevant to EGEE Defining e-Infrastructure for the masses - OGF 25 17

18. Enabling Grids for E-sciencE EGEE-III INFSO-RI-222667 Programming Interface SAGA (GFD 90) Programming Interface DRMAA (GFD 22/133) OGF Compute-Related Standards Defining e-Infrastructure for the masses - OGF 25 Job Management OGSA-BES (GFD 108) HPC Domain Specific Profile HPC Basic Profile (GFD 114) Architecture OGSA EMS Scenarios (GFD 106) Use Cases Grid Scheduling Use Cases (GFD 64) Education ISV Primer (GFD 141) Agreement WS-Agreement (GFD 107) Programming Interface DRMAA (GFD 22/133) Programming Interface SAGA (GFD 90) Accounting Usage Record (GFD 98) Information GLUE Schema Job Definition File Transfer HPC File Staging (GFD 135) Job Description JSDL (GFD 56/136) Application Description HPC Application (GFD 111) Application Description SPMD Application (GFD 115) Job Parameterization Parameter Sweep (Draft) Extend Uses Produces Describes Supports Profiles Production Grid Infrastructure (PGI) WG Profiling JSDL (Wed PM – Leopardi & Thurs AM Dante) PGI-WG Profiling BES (Wed PM – Leopardi & Thurs AM Dante) 18

19. Enabling Grids for E-sciencE EGEE-III INFSO-RI-222667 Access Byte-I/O (GFD 87/88) Naming RNS (GFD 101) OGF Data-Related Standards Defining e-Infrastructure for the masses - OGF 25 Unstructured Data (Disk, Tape) Structured Data (Database) Storage Management SRM (GFD 129) Access Byte-I/O (GFD 87/88) Transport GridFTP (GFD 20) Access & Management DAIS (GFD 74,75,76) Naming RNS (GFD 101) Movement DMI (GFD 134) Data Format Description Language Information Dissemination InfoD (GFD 110) Storage Network Community Existing Possible PGI - Data (Wed AM – Leopardi) 19

20. Enabling Grids for E-sciencE EGEE-III INFSO-RI-222667 Putting all this into context… Defining e-Infrastructure for the masses - OGF 25 Middleware Applications Base Environment (Operating System) Core Middleware Applications Base Environment (Operating System) Community Services Core Middleware Applications Base Environment (Operating System + Grid Services) Community Services 20

21. Enabling Grids for E-sciencE EGEE-III INFSO-RI-222667 Summary The first innovation phase of e-infrastructure is over –The EDG and EGEE series of projects are coming to an end Moving through the standardisation phase –But ‘the only colour you can have is black’! EGI is start of the e-infrastructure exploitation phase –A defined minimal foundation that can be built upon –Enable further innovation, standardisation & exploitation Generating collaborative specifications is hard work –OGF adds negligible overhead to this collaboration OGF’s portfolio of standards is continuing to expand –EGEE’s and its collaborators are driving those critical to us! –Resulting specifications have greater value than having been developed in private Defining e-Infrastructure for the masses - OGF 25 21

22. Enabling Grids for E-sciencE EGEE-III INFSO-RI-222667 Defining e-Infrastructure for the masses - OGF 25 Acknowledgements The GLUE team & Laurence Field Those working on standards in EGEE (& other projects) The e-infrastructure user community within OGF Further standards related information: –http://www.ogf.org/standardshttp://www.ogf.org/standards  Overview of the OGF standards area –http://www.ogf.org/http://www.ogf.org/  Follow the documents link for the detailed technical specifications –http://www.gridtalk.org/http://www.gridtalk.org/  Jargon free ‘Grid Briefings’ available under Documents 22