high resolution regional climate downscaled data transfer...

1
ENERGY ENERGY U.S. DEPARTMENT OF Argonne National Laboratory is a U.S. Department of Energy laboratory managed by U Chicago Argonne, LLC. High resolution regional climate downscaled data transfer using Globus Jiali Wang 1* , V. Rao Kotamarthi 1 , Gagan Agrawal 2 , Raj Kettimuthu 3 , Roee Ebenstein 2 , Yu Su 2 , Torrance I. Leggett 4 1 Environmental Science Division, Argonne National Laboratory 2 Department of Computer Science and Engineering, The Ohio State University 3 Mathematics and Computer Science Division, Argonne National Laboratory 4 Computing, Environment and Life Sciences, Argonne National Laboratory * Contact: [email protected] v Why we need a tool § Pre-process § Transfer between locations § Back-up model outputs § Share with collaborators § Publish to the world § Download by users v Challenges § Transfer speed/performance § Low cost § Security v Machines § Fusion, Argonne § BG/Q (Mira), Argonne § NERSC, Lawrence Berkeley § Any machines used by users v Data size § >200TB, 2TB per model year § Historical/future periods § Different emission scenarios v Globus § Enable intensive analysis § ~10x improvement over sftp § Minimum firewall requirements § No need to VPN/on-site login § Mass storage (HPSS) v SDQuery DSI plug-in § Flexible server-side data subset § Queries based on dimensions/ coordinates/values § Allow users directly perform particular analysis § Support NetCDF and HDF Find the publication here by Su et al. 2013. South-to-north Correlations 2.0 apart 4.0 apart PRISM WRF-PRISM NCEP-PRISM v Climate model data § Why high resolution? § Future climate changes? 1–10 mm 10–20 mm 20–40 mm Acknowledgement: Strategic Environmental Research and Development Program; Argonne Leadership Computing Facility; Argonne Laboratory Computing Resource Center; National Energy Research Scientific Computing Center.

Upload: others

Post on 26-Aug-2020

0 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: High resolution regional climate downscaled data transfer ...kettimut/publications/GlobusWorld_2015_… · GlobusWorld_2015.ppt Author: RajK Created Date: 12/19/2015 11:55:00 PM

ENERGYENERGYU.S. DEPARTMENT OF

Argonne National Laboratory is a U.S. Department of Energy laboratory managed by U Chicago Argonne, LLC.

High resolution regional climate downscaled data transfer using Globus

Jiali Wang1*, V. Rao Kotamarthi1, Gagan Agrawal2, Raj Kettimuthu3, Roee Ebenstein2, Yu Su2, Torrance I. Leggett4 1 Environmental Science Division, Argonne National Laboratory 2 Department of Computer Science and Engineering, The Ohio State University 3 Mathematics and Computer Science Division, Argonne National Laboratory 4 Computing, Environment and Life Sciences, Argonne National Laboratory *Contact: [email protected]

v Why we need a tool §  Pre-process §  Transfer between locations §  Back-up model outputs §  Share with collaborators §  Publish to the world §  Download by users

v Challenges §  Transfer speed/performance §  Low cost §  Security v Machines §  Fusion, Argonne §  BG/Q (Mira), Argonne §  NERSC, Lawrence Berkeley §  Any machines used by users

v Data size §  >200TB, 2TB per model year §  Historical/future periods §  Different emission scenarios

v Globus §  Enable intensive analysis §  ~10x improvement over sftp §  Minimum firewall requirements §  No need to VPN/on-site login §  Mass storage (HPSS) v SDQuery DSI plug-in §  Flexible server-side data subset §  Queries based on dimensions/

coordinates/values §  Allow users directly perform

particular analysis §  Support NetCDF and HDF

Find the publication here by Su et al. 2013.

South-to-north Correlations West-to-east Correlations 2.0 apart 4.0 apart 2.0 apart 4.0 apart

PRIS

M

WRF

-PRI

SM

NCE

P-PR

ISM

WRF

-PRI

SM

NCE

P-PR

ISM

v Climate model data §  Why high resolution?

§  Future climate changes? 1–10 mm 10–20 mm 20–40 mm

Acknowledgement: Strategic Environmental Research and Development Program; Argonne Leadership Computing Facility; Argonne Laboratory Computing Resource Center; National Energy Research Scientific Computing Center.