1. Bob Jones – Project Architecture - 1 March 2002 - n° 1 Project Architecture, Middleware and Delivery Schedule Bob Jones Technical Coordinator, WP12, CERN Bob.Jones@cern.ch

2. Bob Jones – Project Architecture - 1 March 2002 - n° 2 Outline u First year objectives u Architecture overview u Details of each middleware Work Package u Job submission example u Architecture issues and actions u Interaction with PPDG/GriPhyN/iVDGL u Interaction with Globus & Globus Grid Forum (GGF) u Plans for 2002 u Summary

3. Bob Jones – Project Architecture - 1 March 2002 - n° 3 Objectives for the first year of the project u Collect requirements for middleware n Take into account requirements from application groups u Survey current technology n For all middleware u Core Services testbed n Testbed 0: Globus (no EDG middleware) u First Grid testbed release n Testbed 1: first release of EDG middleware u WP1: workload n Job resource specification & scheduling u WP2: data management n Data access, migration & replication u WP3: grid monitoring services n Monitoring infrastructure, directories & presentation tools u WP4: fabric management n Framework for fabric configuration management & automatic sw installation u WP5: mass storage management n Common interface for Mass Storage Sys. u WP7: network services n Network services and monitoring

4. Bob Jones – Project Architecture - 1 March 2002 - n° 4 DataGrid Architecture Collective Services Information & Monitoring Replica Manager Grid Scheduler Local Application Local Database Underlying Grid Services Computing Element Services Authorization Authentication and Accounting Replica Catalog Storage Element Services SQL Database Services Fabric services Configuration Management Configuration Management Node Installation & Management Node Installation & Management Monitoring and Fault Tolerance Monitoring and Fault Tolerance Resource Management Fabric Storage Management Fabric Storage Management Grid Fabric Local Computing Grid Grid Application Layer Data Management Job Management Metadata Management Object to File Mapping Service Index

5. Bob Jones – Project Architecture - 1 March 2002 - n° 5 EDG Interfaces Collective Services Information & Monitoring Replica Manager Grid Scheduler Local Application Local Database Underlying Grid Services Computing Element Services Authorization Authentication and Accounting Replica Catalog Storage Element Services SQL Database Services Fabric services Configuration Management Configuration Management Node Installation & Management Node Installation & Management Monitoring and Fault Tolerance Monitoring and Fault Tolerance Resource Management Fabric Storage Management Fabric Storage Management Grid Application Layer Data Management Job Management Metadata Management Object to File Mapping Service Index Computing Elements System Managers Scientists Operating Systems File Systems Storage Elements Mass Storage Systems HPSS, Castor User Accounts Certificate Authorities Application Developers Batch Systems

6. Bob Jones – Project Architecture - 1 March 2002 - n° 6 WP1: WorkLoad Management u Achievements n Analysis of work-load management system requirements & survey of existing mature implementations Globus & Condor (D1.1) n Definition of architecture for scheduling & res. mgmt. (D1.2) n Development of "super scheduling" component using application data and computing elements requirements u Issues n Distributed nature of WP1 development groups Components Job Description Language Resource Broker Job Submission Service Information Index User Interface Logging & Bookkeeping Service Collective Services Information & Monitoring Replica Manager Grid Scheduler Local Application Local Database Underlying Grid Services Computing Element Services Authorization Authentication and Accounting Replica Catalog Storage Element Services SQL Database Services Fabric services Configuration Management Configuration Management Node Installation & Management Node Installation & Management Monitoring and Fault Tolerance Monitoring and Fault Tolerance Resource Management Fabric Storage Management Fabric Storage Management Grid Application Layer Data Management Job Management Metadata Management Object to File Mapping Service Index

7. Bob Jones – Project Architecture - 1 March 2002 - n° 7 WP2: Data Management u Achievements n Survey of existing tools and technologies data access and mass storage systems (D2.1) n Definition of architecture for data management (D2.2) n Close collaboration with Globus, PPDG/GriPhyN & Condor n Working with GGF on standards n Deployment of GDMP in testbed1 u Issues n Security: clear methods handling authentication and authorization Components SpitFire GDMP Replica Catalog Collective Services Information & Monitoring Replica Manager Grid Scheduler Local Application Local Database Underlying Grid Services Computing Element Services Authorization Authentication and Accounting Replica Catalog Storage Element Services SQL Database Services Fabric services Configuration Management Configuration Management Node Installation & Management Node Installation & Management Monitoring and Fault Tolerance Monitoring and Fault Tolerance Resource Management Fabric Storage Management Fabric Storage Management Grid Application Layer Data Management Job Management Metadata Management Object to File Mapping Service Index

8. Bob Jones – Project Architecture - 1 March 2002 - n° 8 WP3: Grid Monitoring Services u Achievements n Survey of current technologies for info & monitoring in grid environments (D3.1) n Coordination of schemas in testbed 1 n Development of Ftree caching backend based on OpenLDAP to address shortcoming in MDS v1 n Design of Relational Grid Monitoring Architecture (R-GMA) (D3.2) – to be further developed within the context of GGF n Collaboration with Globus for Ftree and PPDG/GriPhyN for res. discovery and schemas n GRM and PROVE adapted to grid environments to support end-user application monitoring u Issues n R-GMA development Components MDS/Ftree R-GMA GRM/PROVE Collective Services Informati on & Monitorin g Replica Manager Grid Scheduler Local Application Local Database Underlying Grid Services Computi ng Element Services Authorizat ion Authentication and Accounting Replica Catalog Storage Element Services SQL Database Services Fabric services Configuratio n Management Configuratio n Management Node Installation & Management Node Installation & Management Monitoring and Fault Tolerance Monitoring and Fault Tolerance Resource Managemen t Fabric Storage Management Fabric Storage Management Grid Application Layer Data Manageme nt Job Manageme nt Metadata Manageme nt Object to File Mapping Service Index

9. Bob Jones – Project Architecture - 1 March 2002 - n° 9 WP4: Fabric Management u Achievements n Survey of existing tools, techniques and protocols for resource specification, configuration and management, as well as integrated cluster management suites (D4.1) n Defined an agreed architecture for fabric management (D4.2) n Initial implementations deployed at several sites in testbed 1 u Issues n Image Installation and Configuration Cache Manager components remain to be integrated Components LCFG PBS & LSF info providers Imagine installation Config. Cache Mgr Collective Services Information & Monitoring Replica Manager Grid Scheduler Local Application Local Database Underlying Grid Services Computing Element Services Authorization Authentication and Accounting Replica Catalog Storage Element Services SQL Database Services Fabric services Configuration Management Configuration Management Node Installation & Management Node Installation & Management Monitoring and Fault Tolerance Monitoring and Fault Tolerance Resource Management Fabric Storage Management Fabric Storage Management Grid Application Layer Data Management Job Management Metadata Management Object to File Mapping Service Index

10. Bob Jones – Project Architecture - 1 March 2002 - n° 10 WP5: Mass Storage Management u Achievements n Review of Grid data systems, tape and disk storage systems and local filesystems (D5.1) n Definition of Architecture and Design for DataGrid storage Element (D5.2) n Collaboration with Globus on GridFTP/RFIO n Collaboration with PPDG on control API n First attempt at exchanging HSM tapes u Issues n Scope and requirements for storage element n Interworking with other Grids Components SE info providers RFIO MSS staging Collective Services Information & Monitoring Replica Manager Grid Scheduler Local Application Local Database Underlying Grid Services Computing Element Services Authorization Authentication and Accounting Replica Catalog Storage Element Services SQL Database Services Fabric services Configuration Management Configuration Management Node Installation & Management Node Installation & Management Monitoring and Fault Tolerance Monitoring and Fault Tolerance Resource Management Fabric Storage Management Fabric Storage Management Grid Application Layer Data Management Job Management Metadata Management Object to File Mapping Service Index

11. Bob Jones – Project Architecture - 1 March 2002 - n° 11 WP6: TestBed Integration u Achievements n Integration of EDG sw release 1.0 and deployment n Working implementation of multiple VOs & basic security infrastructure n Definition of acceptable usage contracts and creation of Certification Authorities group u Issues n Procedures for software integration n Test plan for software release n Support for production-style usage of the testbed Components Globus packaging & EDG config Build tools End-user documents Collective Services Informati on & Monitorin g Replica Manager Grid Scheduler Local Application Local Database Underlying Grid Services Computi ng Element Services Authorization Authentication and Accounting Replica Catalog Storage Element Services SQL Database Services Fabric services Configuratio n Management Configuratio n Management Node Installation & Management Node Installation & Management Monitoring and Fault Tolerance Monitoring and Fault Tolerance Resource Managemen t Fabric Storage Management Fabric Storage Management Grid Application Layer Data Manageme nt Job Manageme nt Metadata Manageme nt Object to File Mapping Service Index WP6 additions to GlobusGlobus

12. Bob Jones – Project Architecture - 1 March 2002 - n° 12 WP7: Network Services u Achievements n Analysis of network requirements for testbed 1 & study of available network physical infrastructure (D7.1) n Collaboration with Dante & DataTAG n Working with GGF (GHPN) & Globus (monitoring/MDS) n Use of European backbone GEANT since Dec. 2001 n Initial network monitoring architecture defined(D7.2) and first tools deployed in testbed 1 u Issues n Resources for study of security issues n End-to-end performance for applications depend on a complex combination of components Components network monitoring tools: PingER Udpmon Iperf Collective Services Information & Monitoring Replica Manager Grid Scheduler Local Application Local Database Underlying Grid Services Computing Element Services Authorization Authentication and Accounting Replica Catalog Storage Element Services SQL Database Services Fabric services Configuration Management Configuration Management Node Installation & Management Node Installation & Management Monitoring and Fault Tolerance Monitoring and Fault Tolerance Resource Management Fabric Storage Management Fabric Storage Management Grid Application Layer Data Management Job Management Metadata Management Object to File Mapping Service Index

13. Bob Jones – Project Architecture - 1 March 2002 - n° 13 A Job Submission Example UI JDL Logging & Book-keeping ResourceBroker Output “sandbox” Input “sandbox” Job Submission Service StorageElement ComputeElement Brokerinfo Output “sandbox” Input “sandbox”InformationService Job Status ReplicaCatalogue Author. &Authen. Job Submit Job Query Job Status

14. Bob Jones – Project Architecture - 1 March 2002 - n° 14 Architecture Issues and Actions u Some concepts remain vague n e.g. interactive jobs u Some boundaries are unclear n e.g scope/functionality of a Storage Element u Some requirements are not yet addressed n e.g. anonymous users u The various software components are not yet fully integrated n e.g. Storage Element, Computing Elements & Info. Sys. u Short term/ Long term trade-offs u Impact of Open Grid Services Architecture n Forthcoming developments by Globus/IBM/GGF u Convergence with US Grid project activities (PPDG/GriPhyN) u The new architecture group will address these points taking into account our experience from testbed1 and further requirements u Implementation of iterative releases, nightly builds and separation of development testbed from production testbed

15. Bob Jones – Project Architecture - 1 March 2002 - n° 15 u Planned intermediate release schedule n TestBed 1:November 2001 n Release 1.1: January 2002 n Release 1.2: March 2002 n Release 1.3: May 2002 n Release 1.4: July 2002 n TestBed 2: September 2002 u Similar schedule will be made for 2003 u Each release includes n feedback from use of previous release by application groups n planned improvements/extension by middle-ware WPs n use of WP6 software infrastructure n feeds into architecture group Plans for 2002 u Extension of testbed n more users, sites & nodes-per-site n Split testbed into development and production sites n Investigate inter-operability with US grids u Iterative releases up to testbed 2 n incrementally extend functionality provided via each Work Package n better integrate the components n improve stability u Testbed 2 (fall 2002) extra requirements n Interactive jobs n Job partitioning for parallel execution n Advance reservation n Accounting & Query optimization n Security design (D7.6) n... demos

16. Bob Jones – Project Architecture - 1 March 2002 - n° 16 Interaction with PPDG/GriPhyN/iVDGL u Work with dataTAG via the InterGrid to investigate inter- operability of US and EU grids 1. Authentication infrastructure –perform cross organizational authentication 2. Unified service discovery and information infrastructure – discover the existence and configuration of service offered by the testbeds. 3. Data movement infrastructure – move data from storage services operated by one organization to another 4. Authorization services – perform some level of cross organization, community based authorization 5. Computational services – coordinate computation across organizations – to allow submission of jobs in EU to run on US sites and vice versa

17. Bob Jones – Project Architecture - 1 March 2002 - n° 17 Interaction with Globus & GGF u Software Licensing – open source agreements for all components u WP1 n Advance Reservation Infrastructure u WP2 n GDMP joint-development and overlap with plans for future Globus Replica Manager n GridFTP/NetLogger integration u WP3 n MDS/Ftree integration n Relationship between R-GMA, GGF and OGSA u WP4 n Authorization capabilities of the Globus gatekeeper n Use resource mgmt subsystem instead of Globus job manager u WP5 n Thread-safe GSI API n RFIO using GridFTP u WP6 n Packaging n Community Authorization Service u WP7 n Integration of MapCentre with MDS n Network message publication to Info. Services

18. Bob Jones – Project Architecture - 1 March 2002 - n° 18 Summary u Application groups requirements defined and analysed u Extensive survey of relevant technologies completed and used as a basis for EDG developments u First release of the testbed successfully deployed u Excellent collaborative environment developed with key players in Grid arena u Project can be judged by: n level of "buy-in" by the application groups n wide-spread usage of EDG software n number and quality of EDG sw releases n positive influence on developments of GGF standards & Globus toolkit