1. Korea Workshop May 2005 1 Grid Analysis Environment (GAE) (overview) Frank van Lingen (fvlingen@caltech.edu) fvlingen@caltech.edu (on behalf of the GAE group)

2. Korea Workshop May 2005 2 Outline GAE System View Frameworks Early Results Projects

3. Korea Workshop May 2005 3 Goal Provide a transparent environment for a physicist to perform his/her analysis (batch/interactive) in a distributed dynamic environment: Identify your data (Catalogs), submit your (complex) job (Scheduling, Workflow,JDL), get “fair” access to resources (Priority, Accounting), monitor job progress (Monitor, Steering), get the results (Storage, Retrieval), repeat the process and refine results Support data transfers ranging from the (predictable) movement of large scale (simulated) data, to the highly dynamic analysis tasks initiated by rapidly changing teams of scientist

4. Korea Workshop May 2005 4 Constraints The “Acid Test” for Grids; crucial for LHC experiments  Large, diverse, distributed community of users  Support for 100s to 1000s of analysis tasks, shared among dozens of sites  Widely varying task requirements and priorities  Need for priority schemes, robust authentication and security Operates in a resource-limited and policy-constrained global system  Dominated by collaboration policy and strategy (resource usage and priorities)  Requires real-time monitoring; task and workflow tracking; decisions often based on a global system view

5. Korea Workshop May 2005 5 NetworkComputeStorage System View Local Services (High Level Services) Global Services (Domain) Applications Monitoring Development Deployment Support/ Feedback Service Oriented Architecture (Frameworks) (Domain) Portal Testing Resources Multiple System Stages Interface Specifications! Input Not only research, but also: Software Engineering Sociology Software Lifecycle Design

6. Korea Workshop May 2005 6 System View (Details) Domains  Virtual Organization and Role management Service Oriented Architecture  Authorized Access  Access Control Management(groups/individuals)  Discoverable  Protocols (XML-RPC, SOAP,….)  Service Version Management  Frameworks: Clarens, MonALISA... Monitoring  End-to-end monitoring,collecting and disseminating information  Provide Visualization of Monitor Data to Users

7. Korea Workshop May 2005 7 System View (Details) Local Services (Local View)  Local Catalogs, Storage Systems, Task Tracking (Single User Tasks), Policies, Job Submission Global Services (Global View)  Discovery Service, Global Catalogs, Job Tracking (Multiple User Tasks), Policies High Level Services (“Autonomous”)  Acts on monitor data and has global view  Scheduling, Data Transfer, Network Optimization, Tasks Tracking (many users)

8. Korea Workshop May 2005 8 System View (Details) (Domain) Portal  One Stop Shop for Applications/Users to access and Use Grid Resources  Task Tracking (Single User Tasks)  Graphical User Interface  User session logging (provide feedback when failures occur) (Domain) Applications  ORCA/COBRA, IGUANA, PHYSH,….

9. Korea Workshop May 2005 9 Single User View 8 Client Application Discovery Planner/ Scheduler Monitor Information Policy Steering Catalogs Job Submission Storage Management Execution 12 3 4 5 5 6 7 Dataset service 9 Data Transfer Catalogs to select datasets, Catalogs to select datasets, Resource & Application Discovery Resource & Application Discovery Schedulers guide jobs to resources Schedulers guide jobs to resources Policies enable “fair” access to resources Policies enable “fair” access to resources Robust (large size) data (set) transfer Robust (large size) data (set) transfer Feedback to users (e.g. status of their jobs) Feedback to users (e.g. status of their jobs) Crash recovery of components (identify and restart) Crash recovery of components (identify and restart) Provide secure authorized access to resources and services. Provide secure authorized access to resources and services. Thousands of user jobs (multi user environment!)

10. Korea Workshop May 2005 10 Service Oriented View Grid Wide Execution Monitoring Application Execution Catalogs Meta Data Virtual Data replica Scheduler Fully Abstract Planner Fully Concrete Planner Grid Services Web Server Analysis Client - Discovery, - ACL management, - Certificate based access HTTP, SOAP, XML-RPC Clients talk standard protocols to “Grid Services Web Server”, Simple Web service API allows simple or complex analysis clients Typical clients: ROOT, Web Browser, …. Clarens portal hides complexity Key features: Global Scheduler, Catalogs, Monitoring, Grid- wide Execution service.

11. Korea Workshop May 2005 11 Peer-2-Peer View discover publish subscribe “Peer-2-Peer” configuration enhances: robustnessrobustness ScalabilityScalabilityProvide: DiscoveryDiscovery  Services  Software  …… PublicationPublication SubscriptionSubscription No Single point of failure

12. Korea Workshop May 2005 12 Framework (Clarens) Authentication (X509) Access control on Web Services. Remote file access (with access control) Discovery of Web Services and Software Shell service. Shell like access to remote machines (managed by access control lists) Proxy certificate functionality Group management VO and role management Good performance of the Web Service Framework Integration with MonALISA 3 rd party application Client Web server Web serverService Clarens Clarens http/https XML-RPC, SOAP. JavaRMI, JSON RPC, …..

13. Korea Workshop May 2005 13 MonALISA JINI Network Web Services (WS), Applications (App) App SS AppS MonALISA Station Servers Framework (MonALISA) WS App WS App WS (1) Publish (2) Disseminate (3) Subscribe (predicates) (4) Steer/Retrieve Monitor Sensors (Web Services/Applications/…) Active filter agents  Process data  Application specific monitoring Mobile agents  decision support  global optimisations MonALISA able to dynamically  register & discover Based on multi-threaded engine  Very scalable Services are self describing Code updates  Automatic & secure Dynamic configuration for services  Secure Admin Interface Fully distributed, no single point of failure!

14. Korea Workshop May 2005 14 GAE Related Projects DISUN (deployment)  Deployment and Support for Distributed Scientific Analysis Ultralight (development)  Treating the network as resource  “Vertically” Integrated Monitor Information  Multi User, resource constraint view MCPS (development)  Provide Clarens based Web Services for batch analysis (workflow) SPHINX (development)  Policy based scheduling (global service) exposed a Clarens Web Service using MonALISA monitor information SRM/Dcache (development)  Service based data transfer (local service) Lambda Station (development)  Authorized programmability of routers using MonALISA & Clarens PHYSH  Clarens based services for command line user analysis CRAB  Client to support user analysis using Clarens Framework

15. Korea Workshop May 2005 15 GAE and Ultralight Make the Network an Integrated Managed Resource Unpredictable multi user analysis Overall demand typically fills the capacity of the resources Real time monitor systems for networks, storage, computing resources,… : E2E monitoring Network Resources Network Planning Request Planning Monitor Application Interfaces Support data transfers ranging from the (predictable) movement of large scale (simulated and real) data, to the highly dynamic analysis tasks initiated by rapidly changing teams of scientists

16. Korea Workshop May 2005 16 Combining Grid Projects into Grid Analysis Environment DISUN Ultralight MCPS Clarens_Applications MonALISA_Applications SPHINX SRM/dCache Lambda Station Policy Grid Analysis Environment Frameworks …….. PHEDEX CRAB PHYSH Development Deployment Support/ Feedback Testing GAE focuses on integration

17. Korea Workshop May 2005 17 GAE Deployment Installation of CMS (ORCA, COBRA, IGUANA,…) and LCG (POOL, SEAL,…) software on Caltech GAE testbed. Serves as environment to integrate applications as web services into the Clarens framework. Demonstrated distributed multi user GAE prototype at SC03, SC04 (using BOSS and SPHINX) Analysis Prototype in January 2005 currently upgraded to work with CRAB. PHEDEX deployed at Caltech, UFL, UCSD and transferring data

18. Korea Workshop May 2005 18 GAE Deployment Clarens has been deployed on ~30+ machines. Other sites: Caltech, Florida, Fermilab, CERN, Pakistan, INFN (see Clarens Discovery Service)  Available in Java and Python Core System: Discovery, Proxy, Authentication, Remote File Access, Shell Service,…. Clarens Discovery Service part of OSG (uses MonALISA)  Talking with EGEE on common interface  Software Discovery Service being developed (SCRAM based prototype available. GAE distributed test framework (Java, Python) available. Daily testing and verification. JavaPythonJavaPython Generic Catalog Service developed to expose pooldbs, refdb, phedex, …and other DBs as entities with key/values

19. Korea Workshop May 2005 19 GAE Deployment PHEDEX and Pubdb Web Services deployed in Florida and Caltech.  Part of the test framework Working with the MCPS group to develop an Clarens based MCPS Web Service interface and browser based GUI.  Supporting Production and Analysis Web Service interface to BOSS (CMS Job/Task book-keeping system) SPHINX: Distributed scheduler developed at UFL Clarens/MonALISA Integration: Facilitating user-level Job/Task Interaction  First version of a Steering Service Java Webstart dashboard being developed for Clarens based services  Also: Cross browser support for JavaScript browser GUI (Firefox, IE, Safari, ….) based on JSON CAVES: Analysis code-sharing environment developed at UFL Work with CERN to have the GAE components included in CMS software distribution. GAE components being integrated in the DPE and VDT distribution used in US-CMS.

20. Korea Workshop May 2005 20 Lessons learned Quality of (the) service(s)  Lot of exception handling needed for robust services (gracefully failure of services)  Time outs are important Need very good performance for composite services Discovery services  enables location independent service composition.  semantics of services are important (different name, name space, and/or WSDL) Web service design: Not every application is developed with a web service interface in mind Interfaces of 3rd party applications change: Rapid Application Development Overlapping functionality of applications (but not same interfaces!) Not one single solution for CMS Not every problem has a technical solution, conventions also important

21. Korea Workshop May 2005 21 -Data movement using PHEDEX -Integration of runjob into current deployment of services  Full chain of end to end analysis -Develop/deploy accounting service -Improved GUI interface (Webstart client) -Improve exception handling -Integrate/interoperability mass storage (e.g. SRM) applications into/with Clarens environment -Steering service -Autonomous replication -Trend analysis using monitor data -E2E error trapping and diagnosis: cause and effect -Strategic Workflow re-planning -Adaptive steering and optimization algorithms -Multi user distributed environment Future Directions

22. Korea Workshop May 2005 22  More Information GAE: http://ultralight.caltech.edu/gaeweb/portal http://ultralight.caltech.edu/gaeweb/portal Ultralight: http://ultralight.caltech.edu http://ultralight.caltech.edu WIKI: http://ultralight.caltech.edu/gaeweb/wiki/ http://ultralight.caltech.edu/gaeweb/wiki/ Thank You. Questions?