Needs for Pacific Regional Cyberinfrastructure

Dr. Sean Cleveland

seanbc@hawaii.edu

Associate Director of Cyberinfrastructure

​

2023-04-12

datascience.hawaii.edu

Topics

• Who are we?

​

• Major Research Drivers for Cyberinfrastructure

​

• How we currently Support Research with CI

​

• Wrap up of Challenges and Future Plans

ITS Cyberinfrastructure Team

Sean Cleveland, PhD – Assoc. Director – Cyberinfrastructure Research Scientist

Ron Merrill, PhD - High Performance Computing

David Schanzenbach - Lead Software Architect

Maria Dumanlang - Communications and Graphic Design

Michelle Choe - Program Manager

Shivana Tanaka - Manager of Student Research Programs

Jennifer Geis -Cyberinfrastructure Software Engineer, Data Engineer

Jared McLean - Research Software Engineer, Data Visualization Scientist

Jeff Wong – Research Software Engineer, Data Scientist

Alan Tsang - Research Software Engineer, Data Scientist

2-6 Graduate & Undergraduate students

Gwen A Jacobs, PhD - Director of Cyberinfrastructure

Mission

Support data intensive research and scholarship at UH with state of the art resources, services and expertise.

​

Our focus is:

• Advanced Research Computing (above the desktop)
• Data
• Science as a Service
• Collaborative research

​

We Serve The Entire UH System

Since 2014

​

Major Drivers For Cyberinfrastructure

HAWAI‘I DATA SCIENCE INSTITUTE

A UH Systemwide initiative to catalyze research and training capacity in data science, computation and visualization.

datascience.hawaii.edu

Data Science Fellows

`Ike Wai Scholars

DATA SCIENCE FELLOWS

• Year Long Fellowship
• Data Science Skill Mastery
• Mentored Research Project

• Professional Development
• Industry Internships
• MS and PhD Students

• 1 month summer program
• Data Science, Analytics, Visualization
• Undergraduate Students

Data Science Fellows

`Ike Wai Scholars

FACULTY, TRAINING AND EDUCATION

• Several new faculty across disciplines
• ML/DL and AI
• Pushing needs for working with Larger Data
• Needing more compute and more powerful

​

• Facilitate the development of data science curricula to serve constituents at all levels
• Data Science Certificate - Information and Computer Sciences - Manoa
• Data Science Certificate & Undergraduate degree – UH Hilo Data Science
• BS in Data Science - Chaminade
• Immersive training in data science and CI skills
• Open to the UH community at large
• Examples: HPC, Cloud computing, Data Wrangling, Python, R, Visualization, Computational workflows

​

​

datascience.hawaii.edu

Hawaiʻi faces daunting challenges as climate change impacts resource availability, ecological sustainability, economic vitality and human health in the islands. To face these challenges, Change Hawaiʻi (HI) is committed to producing actionable science resulting from interdisciplinary data and climate research.

Project Goal

Harness the data revolution to confront climate change and support the state’s transition to a digitized smart economy.

Our Vision

The Change HI project envisions Hawaiʻi with an academic research enterprise that meets the existential threat of climate change with actionable science for stakeholders and meets the training and workforce needs of a diversified and regenerative economy

NSF #OIA-2149133

Data

Monitoring

Data

Processing

Data

Access

Data

Products

Data

Storage

Data

Analysis

Data

Dissemination

The Data Lifecycle

Climate Scientists

Data Scientists

Hydrological Scientists

Ecological

Scientists

Social Scientists

Environmental

Specialists

Tom

Giambelluca

Matt

Lucas

Keri

Kodama

Ryan

Longman

Abby

Frazier

Han

Tseng

Sean

Cleveland

Jared

Mclean

Michael

Dodge II

Jason

Leigh

Roderick

Tabalba

Sayed

Bateni

Yinphan

Tsang

Chris

Shuler

Yu-fen

Huang

Christian

Giardina

Clay

Trauernicht

Elliot

Parsons

Lucas

Berio Fortini

John

Delay

Katie

Kamelamela

Alyssa

Anderson

Melissa

Kunz

Atmospheric

Scientists

Giuseppe

Torri

Allison

Nugent

Cherryle

Heu

Derek

Ford

Patrick

Grady

Emily

Senso

Amy

Schriber

Gwen

Jacobs

Dylan

Giardina

​

• On March 3^rd 2022, the HCDP was officially launched.
• To date, the HCDP has had ~23,000 unique visitors (~40,000 visits)
• Visitors from 115 different countries around the world.
• There have been ~1,000 unique downloads and ~1.6 million files downloaded.

An Infinite number of opportunities!

• Decision support tools
• Hawaiʻi Rangeland Information Portal (HRIP)
• Climate Change Climate Variability and Drought (CCVD) portfolios
• Pacific Drought Knowledge Exchange (PDKE)
• Early-warning
• HI-EMA - Fire risk and warning
• HI-EMA - Flood risk and warning
• USGS - Avian Malaria risk and warning
• Hydrologic modeling
• Improvements to the National Water Model (CIROH)
• Education
• K-12 and beyond

​

Data

Products

Led by UH Mānoa, Asst. Prof. Nicole Hynson and Asst. Prof. Craig Nelson: brings together 50+ researchers across UH Mānoa and Leeward Community College. Goal of establishing a research model for studying the role of microbes in the biology of animals, plants, and the human microbiome in health and disease.

​

Leverage sequencing and metabolomics to infer the composition and activities of microbial communities associated with air, water, soil, insects, and plants as biological characteristics of watersheds.

​

C-MAIKI science gateway to lower the barrier to accessing and running these complex workflows and providing a collaborative data sharing environment. Larger storage capacity allowing larger datasets to support more users and increased sequencing dataset sizes.

​

Cyber-infrastructure and Artificial Intelligence enabled Collaborative Work

UH Mānoa, Prof. Jason Leigh, Asst Prof. Mahdi Belcaid: Smart Amplified Group Environment (SAGE3) (led by Leigh and Belcaid) is NSF-funded scientific software for enabling users to work collaboratively in front of wide-band display environments, to construct artificial intelligence (A) -enhanced data analytics and visualization dashboards that accelerate scientific discovery (NSF #2004014, #2003800, #2003387). SAGE3 enhances support to the community by making it easy for non-AI experts to leverage AI services to conduct data analysis.

Connecting SAGE3 to large external data sources will enhance researchers ability to visualize, collaborate and leverage AI and better support education and training.

Synthetic Aperture Radar Data Collection and Dissemination

UH Mānoa, Asst. Prof. Justin Stopa: Synthetic aperture radar (SAR) acquires high-resolution imagery regardless of cloud cover or daylight on global scales. The open-source European Space Agency (ESA) Sentinel-1 (S-1). The S-1 high-resolution images, called wave mode (WM), are ~20x20 km at 5 m resolution. ~130,000 images are collected each month and the S-1 mission since 2016. The WM archive from 2016 to present exceeds 700TB making it a challenging dataset to manipulate. The SAR images contain a range of geophysical phenomena including rain, atmospheric rolls, atmospheric cells, oceanic fronts, oceanic eddies, and slicks.

Develop image-detection algorithms such as neural networks, and process large datasets, like this one that require fast I/O, will result in new scientific advancements in geophysics and data science.

Coral Reef Mapping

UH Hilo, Asst. Prof. John Burns: Burns’ marine research activities utilize high-resolution 3D mapping of coral reefs, and the data products require substantial computational storage and processing power for rendering and annotating habitat reconstructions.

​

​

​

3D interactive model of Pocillopora meandrina colony from Wai‘opae, Hawai‘i Island (link). Model created by John Burns.

Enable Burn’s lab to run larger analyses. Improve capacity to handle large datasets that are necessary to perform advanced analytical tasks for examining spatial patterns in ecological models.

Rapid ʻŌhiʻa Death (ROD)

UH Hilo – Assoc. Professor Ryan L. Perroy in the Dept of Geography & Env Sci and Director of the Spatial Data Analysis and Visualization (SDAV) research lab

Dr. Perroy’s lab utilizes a suite of geospatial technologies such as small Unmanned Aerial System (sUAS) platforms, LiDAR, and hyperspectral imaging, to support ongoing research and educational projects SDAV is currently monitoring the spread of Rapid ʻŌhiʻa Death (ROD), a strain of fungus that is killing native ʻŌhiʻa trees (Metrosideros polymorpha). Dr Perrory’s group collects repeat high-resolution videos/imagery from sUAS platforms to assess progression rates and spatial patterns of tree mortality associated with ROD.

Processing these longitudinal LiDAR and hyperspectral datasets/products and sharing out the raw and processed data. Further, collaborations with other pacific islands, such as the Marshall Islands & Guam (Dr Romina King >500TB), in receiving and processing similar data that are challenging to transfer without a robust and fault tolerant data transfer infrastructure.

​

So Many More Use Cases!!

How are we currently supporting these researchers?

Data Science Fellows

`Ike Wai Scholars

MANA ADVANCE RESEARCH COMPUTING

• 359 Compute Nodes
• 120 GPUs
• 8,964 cores
• 62.28 TB of memory
• 1 PB of Working research storage
• 61 TB of Flash scratch storage
• 50 TB of Standard scratch storage

$1.8 Million institutional investment 2014,

$700,000 NSF MRI 2019, $400K & $500K NSF CC* 2022

$1.4 Million - 140 Condo Nodes from Researchers

​

​

datascience.hawaii.edu

FREE!!! TO UH Faculty/Researchers/Students/Staff!!!!!

$4.9M

Total Investment in HPC

$1.6M

Data Science Fellows

`Ike Wai Scholars

Koa & KoaStore

• 30 more GPUs - 10 A30 & 20 A400
• 720 more high speed CPU cores
• 750TB more scratch storage and longer work time
• 5-6 more PBs of high performance storage (up to 9x faster)

​

Koa $400K & KoaStore $500K NSF CC* 2022 Awards

Koa FALL 2023 KoaStore SPRING 2024

​

​

datascience.hawaii.edu

FREE!!! TO UH Faculty/Researchers/Students/Staff!!!!!

Data Science Fellows

`Ike Wai Scholars

MANA -

Beyond Your Laptop/Workstation

• Access via the Web Browser or Terminal
• Supports interactive computing for Jupyter/Rstudio/VNC and short compute jobs
• Batch computing for large or long running compute jobs
• High Speed data transfer- 100 Gbps Data Transfer Nodes
• Enabling access to multiple GPUs for Deep Learning and AI

datascience.hawaii.edu

Help With Software/Tools on Mana

• Central Software Repository - modules - installed for everyone
• Custom computing software/environments with Anaconda
• Compiling software - we have intel and open source compilers and knowledge/expertise in helping compile most software

Help With Research Data Movement

• Trouble Moving Data?
• Data Transfers Slow?
• Need to move a lot of Data?

Data Science Fellows

`Ike Wai Scholars

High Speed Data Transfer/Sharing

• UH System has a subscription - so free to researchers
• Very Fast and Robust data transfer - can resume large transfers
• Can transfer from laptop/workstation to Mana, or other major resources NCAR/ACCESS etc
• Can securely share files/folder on your laptop with other Globus users
• Encrypted transfers

​

​

Data Science Fellows

`Ike Wai Scholars

Help Accessing National Resources

​

​

• National NSF cloud computing - Jetstream2
• UH has a regional portion of the Jetstream2 Cloud
• https://jetstream-cloud.org/

​

​

• National HPC resources - ACCESS
• We help with initial access and getting allocations to these resources and helping researcher get their codes running
• https://access-ci.org/

Data Science Fellows

`Ike Wai Scholars

Data Science Fellows

`Ike Wai Scholars

Help With Workflows & Pipelines

• Science-As-A-Service Platform
• Advanced Computing end-to-end workflows
• Functions-As-A-Service (serverless/lambda)
• Streaming Data
• Collaboration, Data Management and Sharing

Example - C-MAIKI Gateway for Microbiome data analysis

​

> 1300 workflows run

>900K parallel jobs on Mana

Simple Browser Access

​

Data Science Fellows

`Ike Wai Scholars

Help With Software & Data Science

• Software Engineering and Data Science Services To Accelerate Your Research
• Access Professional Research Software Engineers & Data Scientists
• Access Graduate Data Fellows

​

​

State of Hawaii Behavioral Health Dashboard

https://bh808.hawaii.gov/

Hawaii Climate Data Portal

https://hawaii.edu/hcdp

Current Projects

Data Science Fellows

`Ike Wai Scholars

Help With Reproducibility

• Looking at your results months/years later for publication or needing to re-run analysis or re-use methods is challenging - set yourself up for success
• Scientific container environments - combining dependencies, data, code and results aides in reproducibility and re-use - especially when combined with scientific computational notebook technologies (jupyter,rstudio)

Data Science Fellows

`Ike Wai Scholars

Help With Data Management

• Funding Agencies and institutions spend $$$$ on research - the outputs (data) are valuable products
• Making data FAIR:
• Findable
• Accessible
• Interoperable
• Re-usable
• We help with grant Data Management Plans

Data Science Fellows

`Ike Wai Scholars

CURRENT FUNDING & COLLABORATIONS

Current Challenges Summary

• More Faculty & Researchers coming & Current users needing more

​

• Small CI staff at UH and other institutions have even less

​

• Storage needs are increasing (Google Drive Unlimited is going away)

​

• Funding to increase support and add more resources

​

• Restrictions on Power/Cooling and Space in Data Centers

Looking for Opportunities to Bring More CI Resources/Capacity to the Region

More Storage

More Compute (GPUs)

​

Looking to Leverage:

OSG, OSDF, NRP, ACCESS, JS2 and NSDF