Tony Doyle Inter Grid Meeting 7 Oct 2001

Tony Doyle Inter. Grid Meeting 7 Oct 2001 Tony Doyle - University of Glasgow

Grid. PP Status § § § § Financial Background Deliverables Recruitment Regional Centres Information Flow Technical Structure Project management Board Tony Doyle - University of Glasgow

£ 17 m 3 -Year Project J 2. 6% I: £ 2. 49 m £ 1. 2 m Experiment Objectives CERN £ 7. 1 m £ 6. 7 m £ 6. 0 m Software Support H: 3. 2% 90. 0% H*: 5. 4% G: Prototype Grid 9. 7% UK . 9% C B C* % % 1. 5 6% 0. B* 1. 4% 1. 5 % D £ 3. 2 £ 2. 9 £ 2. 45 m A 0. 4% E F* D* rs 1 UK Capital 1% % 2. 7 1. 9 1. 5% F age Work Groups A - F 1. 7% 1. 9% Man A* Tony Doyle - University of Glasgow

Major Deliverables § Prototype I Mar 2002 § Performance and scalability testing of components of the computing fabric (clusters, disk storage, mass storage M 9 important: system, system installation, system monitoring) using straightforward physics applications. Testing of the job working demonstrators scheduling and data replication software from the first (public spending review) Data. Grid release (M 9). need to ensure key § Prototype II Mar 2003 areas are visible. Prototyping of the integrated local computing fabric, with § emphasis on scaling, reliability and resilience to errors. Performance testing of LHC applications. Distributed HEP and other science application models using the second Data. Grid release (M 21). § Prototype III Mar 2004 § Full scale testing of the LHC computing model with fabric management and Grid management software for Tier-0 and Tier-1 centres, with some Tier-2 components. This will use the software from the final Data. Grid release (M 33). Tony Doyle - University of Glasgow

First Year Finances § § § Funding of PPARC's EU Data. Grid staff commitment Staff to implement initial Grid Testbed in UK Hardware to satisfy Ba. Bar requirements and EU Data. Grid testbed commitment Staff for CERN LHC computing (Staff costs assume Contribution to CERN Tier-0 only 6 months of salary in first year) § § § 15 EU Data. Grid 3 year posts (already committed) 15 Grid. PP 3 year posts Hardware for Ba. Bar Tier-A/LHC Tier-1 15 CERN 3 year posts Hardware for CERN Tier-0 Total £ 0. 4 M £ 0. 9 M £ 0. 5 M £ 0. 3 M £ 2. 5 M Sets a short-term agenda… complete allocation by December Tony Doyle - University of Glasgow

First Year Deliverables Each Workgroup has detailed deliverables. These will be refined each year and build on the successes of the previous year. The Global Objectives for the first year are: § Deliver EU Data. Grid Middleware (First prototype [M 9]) § Running experiments to integrate their data management systems into existing facilities (e. g. mass storage) § Assessment of technological and sociological Grid analysis needs § Experiments refine data models for analyses § Develop tools to allow bulk data transfer § Assess and implement metadata definitions § Develop relationships across multi-Tier structures and countries § Integrate Monte Carlo production tools § Provide experimental software installation kits § LHC experiments start Data Challenges § Feedback assessment of middleware tools Tony Doyle - University of Glasgow

“two legs good three legs better? ” Build on Existing Structures. Identify Common Requirements. Establish Group Identity. Reach Out. ++ UK Tony Doyle - University of Glasgow

Recruitment Timetable § § § 8 th October - EB input deadline (just after GGF 3). 19 th October - TAG assessment document 26 th October - PMB definition of tasks inc. allocation of posts to Working Groups - first review of CERN posts 2 nd November - Tender document 5 -6 th November - Collaboration Meeting 30 th November- Deadline for submission 7 th December - PRSC mtg. 14 th December- summary document submitted to PPARC outcomes communicated to groups - second review of CERN posts Tony Doyle - University of Glasgow

Regional Centres SRIF Infrastructure Local Perspective: Consolidate Research Computing Optimisation of Number of Nodes? 4 Relative size dependent on funding dynamics Global Perspective: V. Basic Grid Skeleton Tony Doyle - University of Glasgow

Information Flow Tony Doyle - University of Glasgow

Grid. PP Team = Basis for Technical Information Flow? Dissemination/ Collaborations (Roger Jones) Software Support (Andrew Mc. Nab) Workload Management (Dave Colling) Experiments (Nick Brook) Information Services and Data Management (Alex Martin) Security (Dave Kelsey) Testbed (Dave Newbold) Monitoring Services (Steve Fisher) Mass Storage and Fabric Management (John Gordon) Networking (Pete Clarke) CERN (Les Robertson) . . . Technical Task Force Tony Doyle - University of Glasgow

Technical Board Structure? Experiments “Testbed” Optimisation of Prototype Grid, Software Support Number of Groups? . . . 3 -6 -9. . . Security, Networking “Site Management” Fabric Management Mass Storage Build KEY: “Grid Management” Workload Management Data Management Information Services Data. Grid Monitoring Services on merge groups naturally. . . Expand experiments’ role via Pilot developments Tony Doyle - University of Glasgow

Demonstrators real world. . . …simulated world We need both. . M 9 + web-based interfaces… Tony Doyle - University of Glasgow

Grid. PP Year Book Tony Doyle - University of Glasgow

Technical Planning Need to monitor progress… 3 -month report = requirement of all Grid. PP posts …next collab meeting (M 11) Tony Doyle - University of Glasgow

Project Management Board § § § Project Leader Deputy Project Leader Chair of the CB Technical Board EU Data. Grid-UK Leader Resource Manager Experiments Chair Experiments Deputy CERN Liaison e-Science Director (in Att) Tony Doyle John Gordon Steve Lloyd Pete Clarke Robin Middleton Dave Britton Nick Brook Roger Barlow Les Robertson Neil Geddes Final PMB structure determined by appointment. . . Working methods determined quickly via 1 day meeting in Sept. . . Tony Doyle - University of Glasgow

PMB Meetings § Frequency - once per week (fixed weekly slot, e. g. 1 pm) § Agenda circulated 3 hours before (10 am) - web page with links § Duration - 1 hour, via multi-site VC, guillotined § Chair - Project Leader § Agenda items sent to PL = heading + proposed outcome § Order of business determined by Chair § Draft minutes later that day § Minutes of all meetings available via web 2 days later § Actions listed as e. g. ¨ No. Description Initiated Who Target Completed ¨ 1. 1 meeting web site Aug 20 TBD Aug 26 § Actions ordered by completed, short-term, long-term (requiring periodic updates) § Review meeting methods formally after 3 months - external CB report? Tony Doyle - University of Glasgow

Summary § § § § Learning to crawl… … quickly First Priority: EU Data. Grid Next Priority: Integration with experiments Need to respond rapidly to external input Implementation of final Grid. PP structure Definition of scope of Collaboration… § § § Medium Term (M 9): Importance of Demonstrators First steps… § § § Proposal (2/4/01) Data. Grid posts (25/5/01) 1 st Collab. Mtg. (23 -24/5/01) Addendum (13/7/01) Expts. Board (NB, 24/7/01) Collab. Board (SL, 25/7/01) PMB (TD, 1/9/01) Technical Board CERN posts (11/01. . ) 2 nd Collab. Mtg. (5 -6/11/01) Grid. PP posts (1/02. . ) Tony Doyle - University of Glasgow