1
Container SI2-SSE: Scaling Up Science on Cyberinfrastructure with the Cooperative Computing Tools Lifemapper w/Makeflow Species Distr. Modeling http://ccl.cse.nd.edu Integrating Containers and Workflows Our Goal: Make it easy to harness multiple kinds of infrastructure for ordinary applications with no special privileges. Typical User Has an existing code or dataset that works well on a local machine, and now wants to run on 1000s of nodes drawn from a campus cluster, a public cloud, and a national resource. • Open source, GPL License, approx 3 releases per year. • Development focus on Linux and Mac. • Automated build and test using HTCondor + Docker • Users on Blue Waters, XSEDE, OSG, CyVerse, Jetstream. • 1/4FTE + two students: release, outreach, and devel focus. • Used in classes at ND, UWEC, and University of Arizona.. • One workshop at ND + several road tutorials per year. HP24stab Simulated w/Work Queue Matthias Wolf et al, Opportunistic Computing with Lobster: Lessons Learned from Scaling up to 25k Non- Dedicated Cores, Computing in High Energy Physics, 2016. Jakob Blomer et al, The Evolution of Global Scale Filesystems for Scientific Software Distribution, IEEE/AIP Computing in Science and Engineering, 17(6), pages 61-71, December, 2015 DOI: 10.1109/MCSE.2015.111 PI: Douglas Thain, University of Notre Dame (NSF-OCI-1642409) Annual Workshop at ND Lobster HEP Workflow ~10K cores over 7 days, 25K cores over 1 day 12,500 species x 15 climate scenarios x 6 experiments x 500 MB per projection = 1.1M jobs, 72TB of output M Worker Worker T T T T T T . . . 177us of simulated time in 18 days using 10K tasks on 4K cores. Task X Y A B OS Image Execute In: Native HTCondor Mesos Docker Singularity Amazon ECS

SI2-SSE: Scaling Up Science on Cyberinfrastructure PI ... · PDF file• Automated build and test using HTCondor + Docker • Users on Blue Waters, XSEDE, OSG, CyVerse, Jetstream

Embed Size (px)

Citation preview

Page 1: SI2-SSE: Scaling Up Science on Cyberinfrastructure PI ... · PDF file• Automated build and test using HTCondor + Docker • Users on Blue Waters, XSEDE, OSG, CyVerse, Jetstream

Container

SI2-SSE: Scaling Up Science on Cyberinfrastructure with the Cooperative Computing Tools

Lifemapper w/Makeflow Species Distr. Modeling

http://ccl.cse.nd.edu

Integrating Containers and Workflows Our Goal: Make it easy to harness multiple kinds of infrastructure for ordinary applications with no special privileges. Typical User Has an existing code or dataset that works well on a local machine, and now wants to run on 1000s of nodes drawn from a campus cluster, a public cloud, and a national resource.

•  Open source, GPL License, approx 3 releases per year. •  Development focus on Linux and Mac. •  Automated build and test using HTCondor + Docker •  Users on Blue Waters, XSEDE, OSG, CyVerse, Jetstream. •  1/4FTE + two students: release, outreach, and devel focus. •  Used in classes at ND, UWEC, and University of Arizona.. •  One workshop at ND + several road tutorials per year.

Discovering chemistry with an ab initio nanoreactor • LEE-PING WANG, • ALEXEY TITOV, • ROBERT MCGIBBON, • FANG LIU, • VIJAY S. PANDE • & TODD J. MARTÍNEZ • Affiliations • Contributions • Corresponding author Journal name:

Nature Chemistry Volume: 6, Pages: 1044–1048 Year published: (2014)

Discovering chemistry with an ab initio nanoreactor • LEE-PING WANG, • ALEXEY TITOV, • ROBERT MCGIBBON, • FANG LIU, • VIJAY S. PANDE • & TODD J. MARTÍNEZ • Affiliations • Contributions • Corresponding author Journal name:

Nature Chemistry Volume: 6, Pages: 1044–1048 Year published: (2014)

HP24stab Simulated w/Work Queue

Matthias Wolf et al, Opportunistic Computing with Lobster: Lessons Learned from Scaling up to 25k Non-Dedicated Cores, Computing in High Energy Physics, 2016.

Jakob Blomer et al, The Evolution of Global Scale Filesystems for Scientific Software Distribution, IEEE/AIP Computing in Science and Engineering, 17(6), pages 61-71, December, 2015 DOI: 10.1109/MCSE.2015.111

PI: Douglas Thain, University of Notre Dame (NSF-OCI-1642409)

Annual Workshop at ND

O( )

/

Lobster HEP Workflow ~10K cores over 7 days, 25K cores over 1 day

12,500 species x 15 climate

scenarios x 6 experiments x 500 MB per projection = 1.1M jobs, 72TB of output

M

Worker Worker

T T T T T T

. . .

177us of simulated time in 18 days using 10K tasks on 4K cores.

Task

X

Y

A

B

OS Image

Execute In: •  Native •  HTCondor •  Mesos •  Docker •  Singularity •  Amazon ECS