running ligo workflows on the osg

14
Running LIGO workflows on the OSG Britta Daudert OSG-Users meeting Fermilab July 26/27 207 Laser Interferometer Gravitational Wave Observatory G070489-00-R

Upload: gareth-sheppard

Post on 30-Dec-2015

31 views

Category:

Documents


1 download

DESCRIPTION

Running LIGO workflows on the OSG. L aser I nterferometer G ravitational Wave O bservatory. Britta Daudert OSG-Users meeting Fermilab July 26/27 207. G070489-00-R. What are Gravitational Waves?. Gravitational waves are ripples in the fabric of space and time. - PowerPoint PPT Presentation

TRANSCRIPT

Running LIGO workflows on the OSG Britta Daudert

OSG-Users meetingFermilab July 26/27 207

Laser Interferometer Gravitational Wave Observatory

G070489-00-R

What are Gravitational Waves?

• Emitted by accelerating masses

• Sources: Compact Binaries, Bursts, Continuous Wave Source, Stochastic Background

Gravitational waves are ripples in the fabric of space and time

G070489-00-R

The Interferometer

• GW decreases distance between test masses in one arm of the L, increasing it in the other

• Distance is measured by bouncing high-power laser light beams between test masses

• L1, H1, H2 detectorsG070489-00-R

Workflow GenerationMetadata Describing

Raw Data

Analysis Parameters

PIPELINE

Glue Pipeline ModulesLALApps

LAL

DAGDAGMan

DAXPegasus

DAG

LDR DataLocation Service

G070489-00-Rldas-grid OSG

H1Data H2 Data L1 Data

Inside The Pipeline

TmpltbankGenerates bank of waveform parameters

InspiralSignal to noise ratio test

Veto testing

H1Trigger H2 Trigger L1 Trigger

TrigbankConverts triggers into template bank

IncaConsistency tests

Event Candidates

Each chunk Of data is analyzed againstEach template

G070489-00-R

Workflow Size

Each observatory produces data

Each chunk of data is

analyzed with each of the templates

A chunk of data takes about 5 minutes to be analyzed

Each template bank job

takes about 5 minutes Templates + Segments

Work

G070489-00-R

Running Workflows on the OSG

DXDax

Dagx

G070489-00-R

The Inspiral Pipeline On UWMilwaukee

Number of Jobs Local

(No DT)

Number of Jobs OSG

(DT)

Run Time

Local

Run Time

OSG

Rescue Dags OSG

Disk

Usage

189 363 0:52 2:39 1 8.1G

374 717 3:18 3:05 1 17G

828 1585 3:07 3:24 1 39G

2167 4117 10:46 15:46 1 96G

3956 7487 16:52 37:14 2 173G

G070489-00-R

Monitoring with monALISA

G070489-00-R

Challenges and Solutions

Large Data Sets

Large Data Sets Long Data Transfer Times

Disk Space Problems

G070489-00-R

Challenges and Solutions

Large Data Sets

Large Data Sets

Compressing data (LIGO)

Compressing data (LIGO)

SRM SRM

G070489-00-R

Challenges and Solutions

Work Flow Design

Work Flow Design Cleanup issues

Disk space problems

G070489-00-R

Challenges and Solutions

Work Flow Design

Work Flow Design

Dynamic Cleanup (Pegasus 2.0)

Dynamic Cleanup (Pegasus 2.0)

Depth First(Condor_G)

Depth First(Condor_G)

G070489-00-R

G070489-00-R