11
1 University of Sheffield University of Sheffield Dan Tovey Dan Tovey Other’ Other’ Experiments’ Experiments’ Requirements Requirements Dan Tovey Dan Tovey University of Sheffield University of Sheffield

‘Other’ Experiments’ Requirements

Embed Size (px)

DESCRIPTION

‘Other’ Experiments’ Requirements. Dan Tovey University of Sheffield. Background. ‘Other’ Experiments defined as non-LHC and non-US-collider. Covers ANTARES, CALICE, CRESST, H1, LC-ABD, MICE, MINOS, NA48, phenoGrid, SNO, UKDMC, ZEUS etc. Bids for funding under GridPP1 unsuccessful. - PowerPoint PPT Presentation

Citation preview

Page 1: ‘Other’ Experiments’ Requirements

11 University of SheffieldUniversity of SheffieldDan ToveyDan Tovey

‘‘Other’ Experiments’ Other’ Experiments’ RequirementsRequirements

Dan ToveyDan Tovey

University of SheffieldUniversity of Sheffield

Page 2: ‘Other’ Experiments’ Requirements

22 University of SheffieldUniversity of SheffieldDan ToveyDan Tovey

BackgroundBackground ‘‘Other’ Experiments defined as non-LHC and non-US-collider.Other’ Experiments defined as non-LHC and non-US-collider. Covers ANTARES, CALICE, CRESST, H1, LC-ABD, MICE, MINOS, Covers ANTARES, CALICE, CRESST, H1, LC-ABD, MICE, MINOS,

NA48, phenoGrid, SNO, UKDMC, ZEUS etc.NA48, phenoGrid, SNO, UKDMC, ZEUS etc. Bids for funding under GridPP1 unsuccessful.Bids for funding under GridPP1 unsuccessful. Since then have fed requirements (non-grid) for e.g. Tier-1/A Since then have fed requirements (non-grid) for e.g. Tier-1/A

resources into GridPP via EB (DRT rep.).resources into GridPP via EB (DRT rep.). Several EoI-style bids contributed to GridPP2 proposal Several EoI-style bids contributed to GridPP2 proposal

(ANTARES, CALICE, MICE, phenoGrid, UKDMC, ZEUS).(ANTARES, CALICE, MICE, phenoGrid, UKDMC, ZEUS). Developed into full Proforma-2 task definitions:Developed into full Proforma-2 task definitions:

• ANTARESANTARES• CALICECALICE• MICEMICE• phenoGridphenoGrid• UKDMCUKDMC

phenoGrid bid successful phenoGrid bid successful 1.0 FTE for 3 years awarded to 1.0 FTE for 3 years awarded to Durham.Durham.

Others rejected in favour of ‘generic GridPP portal’.Others rejected in favour of ‘generic GridPP portal’.

Page 3: ‘Other’ Experiments’ Requirements

33 University of SheffieldUniversity of SheffieldDan ToveyDan Tovey

ANTARESANTARES

Neutrino telescope being built in the Med.Neutrino telescope being built in the Med. 1 Tb/week raw data rate – would benefit from grid data 1 Tb/week raw data rate – would benefit from grid data

processing.processing. Opportunity for evangelism/outreach to new communities.Opportunity for evangelism/outreach to new communities. Task definition Task definition

• interface ANTARES data and calibration databases with existing interface ANTARES data and calibration databases with existing Grid data handling resources;Grid data handling resources;

• establish appropriate authentication schemes (certification, etc.);establish appropriate authentication schemes (certification, etc.);• enable access to reconstructed data by other collaborators;enable access to reconstructed data by other collaborators;• include ANTARES data into fast, global, astrophysics alert include ANTARES data into fast, global, astrophysics alert

systems.systems. Also proposed Grid running of CORSIKA shower Monte Carlo Also proposed Grid running of CORSIKA shower Monte Carlo

(simple executable suitable for generic portal?)(simple executable suitable for generic portal?) Requested 1.0 FTE/yr for 3 years.Requested 1.0 FTE/yr for 3 years.

Page 4: ‘Other’ Experiments’ Requirements

44 University of SheffieldUniversity of SheffieldDan ToveyDan Tovey

CALICECALICE Major project to define and build calorimetry for a future linear Major project to define and build calorimetry for a future linear

collider.collider. 4 Tb data (104 Tb data (1088 events) from 2004-2006 beam tests – analyse and events) from 2004-2006 beam tests – analyse and

simulate (10simulate (1099 events) using Grid. events) using Grid. Proposed milestones/deliverables:Proposed milestones/deliverables:

• Month 3: CALICE VO and work-load management system in place, with Month 3: CALICE VO and work-load management system in place, with acceptance at Tier-1 + at least one Tier-2 site.acceptance at Tier-1 + at least one Tier-2 site.

• Month 6: CALICE GEANT4 Monte Carlo running in EGEE/LCG Grid Month 6: CALICE GEANT4 Monte Carlo running in EGEE/LCG Grid environment.environment.

• Month 15: Preproduction test run of around 10Month 15: Preproduction test run of around 1077 events based on first events based on first beam test parameters.beam test parameters.

• Month 18: Prototype release for book-keeping and user tools.Month 18: Prototype release for book-keeping and user tools.• Month 21: CALICE VO acceptance at all planned Tier-2 sites.Month 21: CALICE VO acceptance at all planned Tier-2 sites.• Month 24: Start production of first large GEANT4 Monte Carlo dataset.Month 24: Start production of first large GEANT4 Monte Carlo dataset.• Month 27: Completed book-keeping and user tools in place to allow Month 27: Completed book-keeping and user tools in place to allow

multiple dataset generation by non-experts.multiple dataset generation by non-experts. Requested 0.5 FTE/yr for 3 years (ICL proposed, shared with Requested 0.5 FTE/yr for 3 years (ICL proposed, shared with

Babar/GANGA).Babar/GANGA).

Page 5: ‘Other’ Experiments’ Requirements

55 University of SheffieldUniversity of SheffieldDan ToveyDan Tovey

MICEMICE Flagship UK-based proof-of-principle experiment for muon Flagship UK-based proof-of-principle experiment for muon

cooling for a neutrino factory / muon collider.cooling for a neutrino factory / muon collider. Identified four areas where e-Science will be key:Identified four areas where e-Science will be key:

• Data distribution from a Tier 0 centre via the GridData distribution from a Tier 0 centre via the Grid Assumed to be RAL in this case!Assumed to be RAL in this case!

• Apparatus monitoring, control and calibration Apparatus monitoring, control and calibration Remote monitoring developed in collaboration with GridCC (Grid-enabled Remote monitoring developed in collaboration with GridCC (Grid-enabled

Remote Instruments with Distributed Control and ComputationRemote Instruments with Distributed Control and Computation) – unique ) – unique in GridPPin GridPP

• Data persistency Data persistency Adopt LCG toolsAdopt LCG tools

• Grid job submission and monitoring Grid job submission and monitoring Same requirements as typical LHC experiment - adopt CMS/LCG toolsSame requirements as typical LHC experiment - adopt CMS/LCG tools

Requested 1.0 FTE/yr for 3 years at Brunel, assisted by existing Requested 1.0 FTE/yr for 3 years at Brunel, assisted by existing effort at ICL and ICL e-Science studentship.effort at ICL and ICL e-Science studentship.

Page 6: ‘Other’ Experiments’ Requirements

66 University of SheffieldUniversity of SheffieldDan ToveyDan Tovey

phenoGridphenoGrid Proposal from UK phenomenologists to GRID-enable HERWIG++.Proposal from UK phenomenologists to GRID-enable HERWIG++. Project will form part of Generators Services sub-project of LCG.Project will form part of Generators Services sub-project of LCG. Also includes training aspect.Also includes training aspect. Selected milestones/deliverables:Selected milestones/deliverables:

• 0-6 months:0-6 months: Familiarisation with ThePEG structure and HERWIG++ Familiarisation with ThePEG structure and HERWIG++ code. Ensuring code can run under all platforms supported by LCG and code. Ensuring code can run under all platforms supported by LCG and LHC experiments. Modifications necessary to ensure program can be LHC experiments. Modifications necessary to ensure program can be used on GRID. Example jobs and scripts to allow simulation jobs to be used on GRID. Example jobs and scripts to allow simulation jobs to be submitted via GRID. Training of HERWIG and phenoGRID collaboration submitted via GRID. Training of HERWIG and phenoGRID collaboration members.members.

• 7-12 months: 7-12 months: Development of program and interfaces so that can be Development of program and interfaces so that can be used with GANGA and in EDG framework.used with GANGA and in EDG framework.

• 13-18 months: 13-18 months: Development of interfaces to LCG services, e.g. POOL Development of interfaces to LCG services, e.g. POOL storage and passing of events to detector simulation. storage and passing of events to detector simulation.

• 19-24 months:19-24 months: Development of tools to allow simulation of large event Development of tools to allow simulation of large event samples using GRID for distributed parallel computing.samples using GRID for distributed parallel computing.

• 25-30 months: 25-30 months: Validation and testing of interfaces and use of Validation and testing of interfaces and use of HERWIG++ via the GRID HERWIG++ via the GRID tuning new physics models in the program. tuning new physics models in the program.

Requested 1.0 FTE/yr for 3 years (Success! - awarded to Durham).Requested 1.0 FTE/yr for 3 years (Success! - awarded to Durham).

Page 7: ‘Other’ Experiments’ Requirements

77 University of SheffieldUniversity of SheffieldDan ToveyDan Tovey

UKDMCUKDMC UK Dark Matter Collaboration detectors increasing in size/total UK Dark Matter Collaboration detectors increasing in size/total

data rate by factor >10 within next year to ~ 1-5 Tb/year.data rate by factor >10 within next year to ~ 1-5 Tb/year. Further factor 10 increase proposed within 3-5 years (ZEPLIN-Further factor 10 increase proposed within 3-5 years (ZEPLIN-

MAX).MAX). Propose to interface existing reconstruction software Propose to interface existing reconstruction software

(ZAP/UNZAP) and G4 simulations to Grid for distributed running (ZAP/UNZAP) and G4 simulations to Grid for distributed running (UK-UCLA-ITEP etc).(UK-UCLA-ITEP etc).

Make use of existing EDG and LCG tools.Make use of existing EDG and LCG tools. Selected deliverables:Selected deliverables:

• Month 12Month 12: Prototype interfaces between G4/FLUKA and : Prototype interfaces between G4/FLUKA and ZAP/UNZAP, and EDG/LCG tools, making full use of existing ZAP/UNZAP, and EDG/LCG tools, making full use of existing GridPP1 developments (e.g. GANGA).GridPP1 developments (e.g. GANGA).

• Month 24Month 24: Fully functioning online UKDMC detector conditions : Fully functioning online UKDMC detector conditions database and Replica Catalogue accessible by prototype UKDMC database and Replica Catalogue accessible by prototype UKDMC software interfaces.software interfaces.

• Month 36Month 36: Fully functioning UKDMC software framework permitting : Fully functioning UKDMC software framework permitting submission of G4, FLUKA and ZAP/UNZAP reconstruction jobs to submission of G4, FLUKA and ZAP/UNZAP reconstruction jobs to multiple grid-enabled Collaboration computing resources.multiple grid-enabled Collaboration computing resources.

Requested 1.0 FTE/yr for 3 years.Requested 1.0 FTE/yr for 3 years.

Page 8: ‘Other’ Experiments’ Requirements

88 University of SheffieldUniversity of SheffieldDan ToveyDan Tovey

ZEUSZEUS Did not submit Proforma-2: info taken from GridPP2 proposal.Did not submit Proforma-2: info taken from GridPP2 proposal. Proposed to grid-enable existing software framework FUNNEL for Proposed to grid-enable existing software framework FUNNEL for

distributed Monte Carlo job submission.distributed Monte Carlo job submission. Also distributed Also distributed physics analysis tool: already use dCache system physics analysis tool: already use dCache system

for data handling and LSF for resource brokerage on 40 node farm for data handling and LSF for resource brokerage on 40 node farm at DESY at DESY extend across ZEUS institutes world-wide. extend across ZEUS institutes world-wide.

Selected milestones: Selected milestones: • 12 months: Commissioned testbed tailored to ZEUS needs;12 months: Commissioned testbed tailored to ZEUS needs;• 18 months: Authentication strategy established;18 months: Authentication strategy established;• 24 months: I/O layers adapted to suit the Funnel system;24 months: I/O layers adapted to suit the Funnel system;• During 3rd year: Development of resource management concepts and During 3rd year: Development of resource management concepts and

monitoring tools;monitoring tools;• 3 years: Roll-out of production version of Grid-enabled Funnel system 3 years: Roll-out of production version of Grid-enabled Funnel system

to the Collaboration.to the Collaboration. Potential problem with interface between existing framework and Potential problem with interface between existing framework and

EDG/LCG tools however EDG/LCG tools however would need careful design. would need careful design. Problem likely to also occur with use of any generic GridPP portal?Problem likely to also occur with use of any generic GridPP portal?

Page 9: ‘Other’ Experiments’ Requirements

99 University of SheffieldUniversity of SheffieldDan ToveyDan Tovey

LC-ABDLC-ABD

Linear Collider Accelerator and Beam Delivery system Linear Collider Accelerator and Beam Delivery system collaboration.collaboration.

Did not submit Proforma-2: info taken from GridPP2 proposal.Did not submit Proforma-2: info taken from GridPP2 proposal. Need to perform high statistics simulations of physical Need to perform high statistics simulations of physical

interactions between matter (spoilers, collimators etc.) and interactions between matter (spoilers, collimators etc.) and beam particles.beam particles.

Propose to give accelerator codes a Grid interface (GUI portal Propose to give accelerator codes a Grid interface (GUI portal or otherwise)or otherwise) • Interface to Resource Broker, Interface to Resource Broker, • Grid-based metadata catalogue for storage of results Grid-based metadata catalogue for storage of results

Page 10: ‘Other’ Experiments’ Requirements

1010 University of SheffieldUniversity of SheffieldDan ToveyDan Tovey

Generic PortalGeneric Portal Result of GridPP2 PRSC process is proposal for development of Result of GridPP2 PRSC process is proposal for development of

generic GridPP portal for ‘other’ experiments / dissemination generic GridPP portal for ‘other’ experiments / dissemination purposes.purposes.

Clearly vital project if GridPP is to reach out beyond limited set of Clearly vital project if GridPP is to reach out beyond limited set of experiments experiments middleware must be of general use to experiments middleware must be of general use to experiments with little or no spare manpower to tailor own code to Grid use.with little or no spare manpower to tailor own code to Grid use.

Possible use-case for simple submission of generic executable Possible use-case for simple submission of generic executable jobs:jobs:• User submits job to web-based GridPP portal, specifying local User submits job to web-based GridPP portal, specifying local

executable, libraries, steering files, input file metadata, desired executable, libraries, steering files, input file metadata, desired location of output data fileslocation of output data files

Requires generic portal.Requires generic portal.• Tool queries experiment-specific grid-based metadata catalogue to Tool queries experiment-specific grid-based metadata catalogue to

find location of input files (SE).find location of input files (SE). Requires simple, flexible generic grid-based metadata catalogue application Requires simple, flexible generic grid-based metadata catalogue application

which can be adapted simply to needs of wide range of experiments.which can be adapted simply to needs of wide range of experiments.• Tool submits jobs to local CE (or more sophisticated scheduling?)Tool submits jobs to local CE (or more sophisticated scheduling?)• Tool copies output data files to specified location and notifies user has Tool copies output data files to specified location and notifies user has

finished (e-mail).finished (e-mail). Problems with interface with existing experiment-specific software Problems with interface with existing experiment-specific software

frameworks (FUNNEL etc.)?frameworks (FUNNEL etc.)?

Page 11: ‘Other’ Experiments’ Requirements

1111 University of SheffieldUniversity of SheffieldDan ToveyDan Tovey

ConclusionsConclusions

So far progress with Grid use by ‘Other’ experiments not good So far progress with Grid use by ‘Other’ experiments not good due to lack of available manpower.due to lack of available manpower.

GridPP2 has not (so far) provided solution to this problem!GridPP2 has not (so far) provided solution to this problem! Opportunities for evangelising some large international Opportunities for evangelising some large international

collaborations through UK involvement lost.collaborations through UK involvement lost. Nevertheless generic GridPP portal project should certainly Nevertheless generic GridPP portal project should certainly

help some, although possibly not all, experiments.help some, although possibly not all, experiments.• Cautious optimism!Cautious optimism!

Caveat: needs of individual ‘other’ experiments need to be Caveat: needs of individual ‘other’ experiments need to be considered carefully + taken into account when defining considered carefully + taken into account when defining functionality functionality if not risk designing generic tool of use to no- if not risk designing generic tool of use to no-one!one!

In any case non-Grid use of central computing facilities (RAL In any case non-Grid use of central computing facilities (RAL and elsewhere) and elsewhere) mustmust still be supported. still be supported.