teragrid science gateways nancy wilkins-diehr [email protected] teragrid 10, august 2-5, 2010 south...

TeraGrid Science Gateways

Nancy [email protected]

TeraGrid 10, August 2-5, 2010

South Tenth Street Bridge, Pittsburgh

TeraGrid resources today include:•Tightly Coupled Distributed Memory Systems, 2 systems in the top 10 at top500.org

– Kraken (NICS): Cray XT5, 99,072 cores, 1.03 Pflop– Ranger (TACC): Sun Constellation, 62,976 cores, 579 Tflop, 123 TB RAM

•Shared Memory Systems– Cobalt (NCSA): Altix, 8 Tflop, 3 TB shared memory– Pople (PSC): Altix, 5 Tflop, 1.5 TB shared memory

•Clusters with Infiniband– Abe (NCSA): 90 Tflops– Lonestar (TACC): 61 Tflops– QueenBee (LONI): 51 Tflops

•Condor Pool (Loosely Coupled)– Purdue- up to 22,000 cpus

•Gateway hosting– Quarry (IU): virtual machine support

•Visualization Resources– TeraDRE (Purdue): 48 node nVIDIA GPUs– Spur (TACC): 32 nVIDIA GPUs

•Storage Resources– GPFS-WAN (SDSC)– Lustre-WAN (IU)– Various archival resources

TeraGrid 10, August 2-5, 2010 Source: Dan Katz, U Chicago

But change is consta

nt - new sy

stems:

•Data Analysis and Vis s

ystems

•Longhorn (TACC): D

ell/NVIDIA, C

PU and GPU

•Nautilus (N

ICS): SGI U

ltraViolet, 1

024 cores,

4TB global shared memory

•Data-Intensiv

e Computing

•Dash (SDSC): I

ntel Nehalem, 544 processo

rs,

4TB flash memory

•FutureGrid

•Experimental computing grid

and cloud test-bed

to tackle research challenges in computer sc

ience

•Keeneland

•Experimental, h

igh-perform

ance computing

system with NVIDIA Tesla

accelerators

So how do Gateways fit into this?Gateways are a natural result of the impact of the internet on

worldwide communication and information retrieval


Only 18 years since the release of Mosaic!

vt100 in the 1980s and alogin window on Ranger today


Why are gateways worth the effort?

•Increasing range of expertise needed to tackle the most challenging scientific problems–How many details do you

want each individual scientist to need to know?•PBS, RSL, Condor•Coupling multi-scale codes•Assembling data from multiple sources

•Collaboration frameworks


#! /bin/sh#PBS -q dque#PBS -l nodes=1:ppn=2 #PBS -l walltime=00:02:00#PBS -o pbs.out#PBS -e pbs.err#PBS -Vcd /users/wilkinsn/tutorial/exercise_3../bin/mcell nmj_recon.main.mdl

+( &(resourceManagerContact="tg-login1.sdsc.teragrid.org/jobmanager-pbs") (executable="/users/birnbaum/tutorial/bin/mcell") (arguments=nmj_recon.main.mdl) (count=128) (hostCount=10) (maxtime=2) (directory="/users/birnbaum/tutorial/exercise_3") (stdout="/users/birnbaum/tutorial/exercise_3/globus.out") (stderr="/users/birnbaum/tutorial/exercise_3/globus.err"))

=======# Full path to executableexecutable=/users/wilkinsn/tutorial/bin/mcell

# Working directory, where Condor-G will write # its output and error files on the local machine.initialdir=/users/wilkinsn/tutorial/exercise_3

# To set the working directory of the remote job, we# specify it in this globus RSL, which will be appended# to the RSL that Condor-G generatesglobusrsl=(directory='/users/wilkinsn/tutorial/exercise_3')

# Arguments to pass to executable.arguments=nmj_recon.main.mdl

# Condor-G can stage the executabletransfer_executable=false

# Specify the globus resource to execute the jobglobusscheduler=tg-login1.sdsc.teragrid.org/jobmanager-pbs

# Condor has multiple universes, but Condor-G always uses globusuniverse=globus

# Files to receive sdout and stderr.output=condor.outerror=condor.err

# Specify the number of copies of the job to submit to the condor queue.queue 1

Gateways democratize access to high end resources

•Almost anyone can investigate scientific questions using high end resources–Not just those in the research groups of those who request

allocations–Gateways allow anyone with a web browser to explore

•Opportunities can be uncovered via google–My then 11-year-old son discovered nanoHUB.org when his science class was studying Bucky Balls

•Foster new ideas, cross-disciplinary approaches–Encourage students to experiment

•But used in production too–Significant number of papers resulting from gateways including

GridChem, nanoHUB–Scientists can focus on challenging science problems rather

than challenging infrastructure problemsTeraGrid 10, August 2-5, 2010

Today, there are approximately 35 gateways using the TeraGrid


Not just ease of useWhat can scientists do that they

couldn’t do previously?•Linked Environments for Atmospheric Discovery (LEAD) - radar data coupled with on demand computing•National Virtual Observatory (NVO) – access to sky surveys•Ocean Observing Initiative (OOI) – access to sensor data•PolarGrid – access to polar ice sheet data•SIDGrid – expensive datasets, analysis tools•GridChem –coupling multiscale codes

•How would this have been done before gateways?TeraGrid 10, August 2-5, 2010

3 steps to connect a gateway to TeraGrid

•Request an allocation–Only a 1 paragraph abstract

required for up to 200k CPU hours

•Register your gateway–Visibility on public TeraGrid page

•Request a community account–Run jobs for others via your

portal

•Staff support is available!•www.teragrid.org/gateways


Tremendous Opportunities Using the Largest Shared Resources -

Challenges too!•What’s different when the resource doesn’t belong just to me?–Resource discovery–Accounting–Security–Proposal-based requests for resources (peer-reviewed access)

•Code scaling and performance numbers•Justification of resources•Gateway citations

•Tremendous benefits at the high end, but even more work for the developers•Potential impact on science is huge

–Small number of developers can impact thousands of scientists

–But need a way to train and fund those developersTeraGrid 10, August 2-5, 2010

When is a gateway appropriate?

•Researchers using defined sets of tools in different ways–Same executables, different input

•GridChem, CHARMM

–Creating multi-scale workflows–Datasets

•Common data formats–National Virtual Observatory–Earth System Grid–Some groups have invested significant efforts here

•caBIG, extensive discussions to develop common terminology and formats

•BIRN, extensive data sharing agreements

•Difficult to access data/advanced workflows–Sensor/radar input

•LEAD, GEON


How to get started?

•Conduct a needs assessment–Should I build a gateway?–Can I use an existing gateway?–What problems am I trying to solve?

•All gateways don’t need high end computing

•Decide on a software approach–Recommended software at www.teragrid.org

•Targeted effort by a few can benefit many–Could a pool of developers design gateways for different

domain areas? Yes!

•TeraGrid staff assistance


Linked Environments for Atmospheric Discovery (LEAD)

•Providing tools that are needed to make accurate predictions of tornados and hurricanes

•Meteorological data•Forecast models•Analysis and visualization tools

•Data exploration and Grid workflow


Highlights: LEAD Inspires StudentsAdvanced capabilities regardless of location

•A student gets excited about what he was able to do with LEAD•“Dr. Sikora:Attached is a display of 2-m T and wind depicting the WRF's interpretation of the coastal front on 14 February 2007. It's interesting that I found an example using IDV that parallels our discussion of mesoscale boundaries in class. It illustrates very nicely the transition to a coastal low and the strong baroclinic zone with a location very similar to Markowski's depiction. I created this image in IDV after running a 5-km WRF run (initialized with NAM output) via the LEAD Portal. This simple 1-level plot is just a precursor of the many capabilities IDV will eventually offer to visualize high-res WRF output. Enjoy!• Eric” (email, March 2007)


Community Climate System Model (CCSM)

•Makes a world-leading, fully coupled climate model easier to use and available to a wide audience•Compose, configure, and submit CCSM simulations to the TeraGrid

•Used in Purdue’s POL 520/EAS 591: Models in Climate Change Science and Policy

–Semester-long projects, 100 year CCSM simulations, generate policy recommendations based on scientific, economic, and political models of climate change impacts


Analytical UltracentrifugationEmerging computational tool for the study of proteins

•The Center for Analytical Ultracentrifugation of Macromolecular Assemblies, UT Health Sciences – Major advances in the

characterization of proteins and protein complexes as a result of new instrumentation and powerful software

– Monitoring the sedimentation of macromolecules in real time in the centrifugal field allows their hydrodynamic and thermodynamic characterization in solution

– Observations are electronically digitized and stored for further mathematical analysis

– http://uslims.uthscsa.edu/TeraGrid 10, August 2-5, 2010

Source: Modern analytical ultracentrifugation in protein science: A tutorial review, Wikipedia

UltraScan provides a comprehensive data analysis environment

•Management of analytical ultracentrifugation data for single users or entire facilities•Support for storage, editing, sharing and analysis of data

–HPC facilities used for 2-D spectrum analysis and genetic algorithm analysis•TeraGrid (~2M CPU hours used)•Technische University of Munich•Juelich Supercomputing Center

•Portable graphical user interface•MySQL database backend for data management•Over 30 active institutions•TeraGrid advanced support

–Fault tolerance, workflows, use of multiple TG resources, community account implementation


Southern California Earthquake Consortium (SCEC) Gateway used to produce realistic

hazard map•Probabilistic Seismic Hazard Analysis (PSHA) map for California– Created from Earthquake

Rupture Forecasts (ERC)•~7000 ruptures can have 415,000 variations

•Warm colors indicate regions with a high probability of experiencing strong ground motion in the next 50 years•Ground motion calculated using full 3-D waveform modeling for improved accuracy– Results in significant CPU use


SCEC: Why a gateway?

•Calculations need to be done for each of the hundreds of thousands of rupture variations–SCEC has developed the “CyberShake computational

platform”•Hardware, software and people which combine to produce a useful scientific result

–For each site of interest - two large-scale MPI calculations and hundreds of thousands of independent post-processing jobs with significant data generation»Jobs aggregated to appear as a single job to the TeraGrid»Workflow throughput optimizations and use of SCEC’s

gateway “platform” reduced time to solution by a factor of three

–Computationally-intensive tasks, plus the need for reduced time to solution is a priority make TeraGrid a good fit


Source: S. Callahan et.al. “Reducing Time-to-Solution Using Distributed High-Throughput Mega-Workflows – Experiences from SCEC CyberShake”.

Social Informatics Data GridCollaborative access to large, complex datasets

•SIDGrid is unique among social science data archive projects–Streaming data which

change over time•Voice, video, images (e.g. fMRI), text, numerical (e.g. heart rate, eye movement)

– Investigate multiple datasets, collected at different time scales, simultaneously•Large data requirements•Sophisticated analysis tools


http://www.ci.uchicago.edu/research/files/sidgrid.mov

Viewing multimodal data like a symphony conductor

•“Music-score” display and synchronized playback of video and audio files– Pitch tracks– Text– Head nods, pause, gesture

references

•Central archive of multi-modal data, annotations, and analyses– Distributed annotation efforts by

multiple researchers working on a common data set•History of updates

•Computational tools– Distributed acoustic analysis using

Praat– Statistical analysis using R– Matrix computations using Matlab

and Octave


Source: Studying Discourse and Dialog with SIDGrid, Levow, 2008

Geographic Information Systems and HPC

•GISolve–Data-intensive, large- and multi-scale spatial analysis and modeling increasingly important for scientific discovery and decision-making•Ecology, environmental sciences, geosciences, public health, and social sciences

–Short demo


Future Technical Areas•Web technologies change fast

– Must be able to adapt quickly

•Gateways and gadgets– Gateway components

incorporated into any social networking page

– 75% of 18 to 24 year-olds have social networking websites

•iPhone apps?•Web 3.0

– Beyond social networking and sharing content

– Standards and querying interfaces to programmatically share data across sites

•Resource Description Framework (RDF), SPARQL


Gateways can further investments in other projects

•Increase access–To instruments, expensive data collections

•Increase capabilities–To analyze data

•Improve workforce development–Can prepare students to function in today’s cross-disciplinary

world

•Increase outreach•Increase public awareness

–Public sees value in investments in large facilities–Pew 2006 study indicates that half of all internet users have been

to a site specializing in science–Those who seek out science information on the internet are more

likely to believe that scientific pursuits have a positive impact on society


But, sustained funding is a problem

•Gateways can be used for the most challenging problems, but–Scientists won’t rely on something that they are not confident

will be around for the duration•We see this with software, but even more so with gateway infrastructure

•A sustained gateway program can–Reduce duplication of effort

•Sporadic development with many small programs

– Increase diversity of end users– Increase skill set diversity of developers–Bring together teams to address the toughest problems


Gateway SustainabilitySmall, non-TG, EAGER grant

•Characteristics of short gateway funding cycles– Build exciting prototypes with input

from scientists– Work with early adopters to extend

capabilities– Tools are publicized, more scientists

interested– Funding ends– Scientists who invested their time to

use new tools are disillusioned•Less likely to try something new again

– Start again on new short-term project

•Need to break this cycle•EAGER grant to look at characteristics of successful gateways and domain areas where a gateway could have a big impact•www.sciencegateways.org

Science Advisory Board, July 19-20, 2010

teragrid science gateways nancy wilkins-diehr [email protected] teragrid 10, august 2-5, 2010 south...

Documents

pbs e

pbs o

orgjobmanagerpbs executable

pbs q dque

teragrid resources

tb shared memoryclusters

working directory

condorg generatesglobusrsl