Applications Review RWL Jones GridPP12. ATLAS Risks: –Delayed middleware, raid changes in framework, competition Progress –New GANGA version (first deliverable)

Applications ReviewApplications Review

RWL JonesGridPP12

ATLASATLAS

• Risks: – Delayed middleware, raid changes in framework, competition

• Progress– New GANGA version (first deliverable) for ATLAS use with:

• plug-in to ATLAS production system through ADA• plug in for ATLAS AMI (metadata system)

– Developing ADA job editor and builder, AMI browser – Support of Grid operations –

• software kits and distribution tools (ADA package management) – support for OS variants– Frederic Brochu as production manager

• will be part of the Grid tool selection ream• Issues

– Interoperability issues between Grids– SE issues

• LCG Castor failures• RLS corruption (transfer and m/w failure).

LHCbLHCb

• Risks: – Technology incompatibility– Changing metadata/prodsys requirements– Dependence on ARDA etc.

• Progress– No level-1s due until the end of 05Q1.– Production console prototype advanced because of changed DC

schedule. – Production desktop release in DIRAC was a level-1 in GridPP1,

now delivered and in use.– Bookkeeping interface developed in ARDA context.

• A Tomcat servelet is deployed • The XML-RPS service development is suspended in favour of the

GANGA/ARDA interface.• Issues

– Changing DC planning – Integration with/dependence on GANGA and ARDA.

GANGA GANGA

• Risks: – Contention between ATLAS and LHCb objectives– Competition– Sudden changes in frameworks– Middleware and metadata delays.

• Progress– Revision of design underway to optimise link to client systems.– Updates to job registry to allow reloading of pre-repackaging jobs. – Improvements in the threading and monitoring of the core modules. – Athena (Tan, for ATLAS) and Brunel (Soroko for LHCb) modules plugged-in

using PyBus. Improvements in the DIRAC and DaVinci application handlers.

– Python bindings for DIAL, access to dial classes using PyDial • Allows access to the ATLAS production system etc.

– A prototype job submission is implemented. – Jobs splitting is being developed and extensive testing underway.

• Issues – Slight delays in level-2s because priority given to experiment-specific

features• expected they will be recovered during Soroko’s visit to CERN. • Similar short trip by Harrison and Tan to BNL (2 weeks) requested to move

things forward.

CMS CMS

• Risks: – Everything! – They list withdrawal of an experiment

• Progress– Significant contribution to PhEDEx – could be of more general

use.• Can sustain TB transfers• Works across LCG and Grid3

– Work towards bookkeeping metadata system – Unification of Italian (Grape) and UK (Gross) job submission.

• BOSS will be re-engineered to include required functionality – R-GMA performance studies indicate it will scale, deployment on

LCG-2 test bed• Testing continues in CMS context

• Issues – They list withdrawal of an experiment as a risk!– Concerns about disparate effort met in large part by central

planning

SAMGrid SAMGrid

• Risks: – Almost everything – They also list withdrawal of an experiment

• Progress– CDF

• Integrating SAM and Grid3 – Is this covered by JIM/RunJob deliverables?

• basic job submission demonstrated. • JIM installation at Oxford and Glasgow. • SAM v6 switch took longer than needed because of memory lean in db-server.

– D0: • Installing and testing SAMGrid at IC to allow P17 reprocessing. • Adding reprocessing to D0RunJob. (Level-2s on track)

– SAM interfaces underway to allow the reprocessing to be of general use.

• Issues – They list withdrawal of an experiment as a risk!– Managerial – need for more clearly defined Level-3s– CDF plans may face major changes after central review– When do the Regional Analysis Centres become live and start reporting?

BaBar BaBar

• Risks: – LCG and other middleware – Gird divergence/interoperability– Breaking dependence on Objectivity

• Progress– Requirement capture for distributed analysis complete.– Grid submission Manchester/RAL, but problems with

conditions db access, environment variables– RLS read-write access to Italy demonstrated

• Issues – Reporting!– Affect of late hiring on delivery

QCDGrid QCDGrid

• Risks: – External middleware– Staff retention – Change of project scope

• Progress– ILDG look likely to adopt QCDgrid software

• C++ + Globus + EDG security– Risk analysis and stress testing requirement capture

complete – Good progress on stress testing itself and software

installation and distribution on UKQCD machines.– QCDOC is now installed and being shaken down in

Edinburgh along with ‘Tier 1’ 50 Tbyte store.• Issues

– Contact/overlap with other GridPP activities

PhenoGrid PhenoGrid

• Risks: – No risk assessment

• Progress– ?

• Issues – Reporting!– Affect of late hiring on delivery

Portal Portal

• Risks: – Beaten to the task by others

• Progress– Requirements complete but technical review

document not on time. • Needs discussion with Rob Allen

– First portlet built allowing GridSphere application access to DN server

• can issue visual_grid_proxy_init using jnlp.

• Issues – Need swift progress– Links with other projects need to be established