Founded by Duke, NC State and UNC

1
Renaissance Computing Institute

• Founded by Duke, NC State and UNC
• Leading edge technologies
• High performance computing, networking, data,
  visualization
• Broad range of technical core expertise and
  competencies
• Facilities, technical resources and expertise
  located across the state
• Central sites at Duke, NC State, UNC and Europa
• Regional Engagement Facilities at ECU,
  UNC-Asheville, UNC-Charlotte

2
RENCI

• Vision/Mission
• RENCI is a multi-institutional organization
  which brings together multidisciplinary experts
  and advanced technological capabilities to
  address pressing research issues and to find
  solutions for complex problems that affect the
  quality of life in North Carolina, our nation and
  the world.
• What we do
• Bring technology and expertise to collaborative
  projects
• Bring together staff from across the state and
  nation to support and develop projects and
  applications
• Act as a convener for multi-institutional and
  multidisciplinary projects and programs
• Serve as the glue and facilitator for
  collaborative projects

Catalyst for Innovation
3
RENCI Collaboration Model
Disaster and Environmental
Arts and Humanities
Public Health
Initiatives and Projects
Collaboration, and Engagement
RENCI Expertise
University Expertise
Advanced IT Technologies
RENCI Technologies
Visualization
Data
Computing
Networking
4
Organization
Oversight Board
Triangle CI Council
Advisory Committee
Domain Scientists Kirk Wilhelmsen - Genomics
Rick Luettich Coastal Science Joyce Rudinsky
Humanities Mary Whitton Computer Sci
Alan Blatecky Interim Director
Triangle Engagement Centers - Duke Marilyn
Lombardi - NCSUTheresa-Marie Rhyne - UNC Ruth
Marinshaw
John Gallagher Business Ops
Charles Schmitt BioSciences
John McGee Cyberinfrastructure
Ken Galluppi Disaster Studies
Ilia Baldine Networking
Karen Green Com Outreach
Rob Fowler HPC CS
David Knowles Econ Engage.
Pat Dreher Infrastructure Ops
Ray Idaszak Visualization
Regional Engagement Sites UNC Asheville, East
Carolina University, UNC Charlotte
5
One network, but many collaboration models
6
NC Floodplain Remapping

• New floodplain maps for FEMA for tropical and
  extra-tropical storms
• RENCI contribution
• Create high-res digital elevation model (DEM)
  from LIDAR and bathymetric data
• Compute flood hazard curves for coastal
  floodplain of NC using ADCIRC

Resulting 10-m DEM
7
Computational Effort

• 324 individual simulations, each with a
  probability of occurrence (Weighting from Joint
  Probability Method)
• Each simulation takes about 12-16 hours on
  RENCIs IBM BlueGene/L
• 256 processors per sim, 10-12 simultaneous
  simulations
• 1.2 million cpu-hours

8
UNC Asheville Flooding

• This Visualization utilizes GIS data to depict a
  model of a flooding scenario in the Biltmore
  Village area of Asheville, NC.
• Flooding in this local area is caused by the
  Swannanoa River, the main tributary of the
  Swannanoa Watershed, breaching its banks and
  entering into its floodplain.
• In this particular scenario, the depicted
  floodwater surface polygon (shown in transparent
  blue) is actual FEMA data based on the 100-year
  floodplain for this area.
• The result helps users better understand what
  flooding will look like in a given area based on
  different scenarios including 5-year, 10-year,
  50-year and even 500-year flood events.

9

RENCI Integrated Desktop Collaboration

• Serving the North Carolina Emergency Management
  Community

10
Desktop Collaboration

• SmileTiger video conferencing is in use around
  the state, with over 200 trained.
• Used around the state and by NC EM management in
  Tropical Storm Hanna
• NCEMA SharePoint Website provides improved
  communications
• The portal to NCFirst, SmileTiger and other
  capabilities.

Web Portal
11
NC-FIRST

• Weather information program for NC emergency
  managers
• Two parts
• Classroom training
• 400 EM/ES personnel trained
• Web portal

12
Automated Vehicle Tracking (AVL)

• Low cost and simple Emergency Vehicle Tracking
• Using Open Source OpenGTS
• No Licensing issues and open for future
  development and integration with other RENCI
  tools
• Multiple viewing and report formats
• Currently testing with Chapel Hill Fire Dept. and
  Lee County Emergency Mgt.

Track
Real time tracking of CHFD Engine 31
13
Virtual Environments for Emergency Response

• Explores emergency response resource management
  and collaboration via a virtual environment
• Live telemetry drives location of resources
  depicted above large map on ground floor
• Each floor of bldg includes virtual environment
  tools to facilitate coordination, communication
  and collaboration among users in different
  geographic locations

14
Virtual Environments for Emergency Response
15
Brunswick County Flood Sensors

• During tropical cyclones, major evacuation routes
  in Brunswick County flood easily
• Police officers are deployed to locations to wait
  for flood
• Goal provide real-time info to officials and NWS
  during storm
• How much rain is falling?
• How fast is the water rising?

16
Brunswick County Flood Sensors

• Work with county officials and NWS to provide
  observations at easily flooded points
• Weather stations (5)
• Water level sensors (eventually 7)

Weather Stations
Water Level Sensors
17
NC EONS - Environmental Observation Network
System for North Carolina
A multi-institutional project building an
instrumented science platform 4.3 miles into the
Pamlico Sound to study marine science and biology
of this ecologically and economically critical
area. Instruments for weather, water currents,
waves, water chemistry, an biology.
18
BEN Overview

• BEN is an experimental dark fiber facility
• Will support experimentation at metro scale
• Distributed applications researchers
• Networking researchers
• Not a production network
• Main goal enabling disruptive technologies
• Shared by the researchers
• Coarse-grained time sharing is the primary mode
  for usage
• Assumes some experiments must be granted
  exclusive access to the infrastructure

19
Examples of Empowering University Scientists

• Providing Access To Computer Power
• Protein Modeling and
• Engineering
• Provide Technology
• 3000x increased speed DNA sequence assembly so
  that Equipment Purchased with UNC Cancer Research
  Fund can Be used more effectively
• Both
• Developed and Coded Algorithm Called CHAT that
  has found important causes of Parkinsons
  Disease, ALS and Breast Cancer

Kulhman Lab
20
CHAT found 9 of 96 unrelated PD patients share
large parts of a 1000 SNP haplotype that spans 9
megabase
(screen clipped to left)
leucine rich repeat and fibronectin type III
domain containing 5 IGcam Immunoglobulin
domain cell adhesion molecule (cam) subfamily
members are components of neural cell adhesion
molecules (N-CAM L1), Fasciclin II and the insect
immune protein Hemolin.
21
North Carolina Collaboratorium for Bioinformatics

• How
• Create the infrastructure to support a culture
  that facilitates data aggregation and re-use that
  protects the interests of participants
• Create a state- wide HIPPA administrative unit
  for data ownership

• What
• A Federated Database of for the State all for
  Biomedical Data
• Virtual Workspace that promotes efficient
  cooperative interdisciplinary research

• Why
• New Research Opportunities
• Increased Competitiveness of State and
  Universities
• Results that can be measured in lives and

22
RENCI_at_UNC Projects
Melanoma Image Analysis
Spectacular Justice
Digital Media Class
Manhattan Wind Fields
23
UNC Campus Network Visualization

• A visualization of UNC campus network traffic
  where each UNC subnet is assigned a geographic
  location.
• When a connection exists between each subnet, an
  arc is drawn from on location to the connecting
  location.
• Each connection is colored based on type web,
  file transfer, etc.
• Enhances insight and understanding of network
  traffic and can be used to identify anomolous
  network behavior.

24
UNC Grant Visualization

• Interactive Visualization of all active UNC
  Chapel Hill grants
• Intuitively allows collaborators to discover what
  geographical and topical areas are getting
  funded and what individuals they might
  collaborate with
• The visualization animates day by day from the
  day the first grant in the database becomes
  active until the last one ends.

25
RENCI_at_NCSU Visualization Projects
Visualization by Sidharth Thakur, Ph.D.
Image by Steve Chall Theresa-Marie Rhyne
Interactive Charts Visualization ot NC Export
data for Textiles products. Created with Dr.
Nancy Cassill her team, NC State Univ. Textile
Apparel Technology Management Department.
Visualization of Data Intense Global Climate
Change Models. Created with Dr. Gary Lackmann
his team, NC State Univ. Marine, Earth
Atmospheric Sciences Dept.
26
RENCI_at_NCSU Visualization Projects
Visualization by Sidharth Thakur, Ph.D.
Visualizaiton by Steve Chall
Visualization of the Change in Net Migration
over a 10 Year period (1990 2000) in percentage
for the state of North Carolina. Created with Dr.
Roland Stephen his colleagues, NC State Univ.
Institute for Emerging Issues.
Visualization of Micro-Mechanical Properties of
Soils to evaluate Geotechnical Engineering
concerns. Created with Dr. Matt Evans his team,
NC State Univ. Civil, Construction
Environmental Engineering Department.
27
Docuverse

• Every documentis represented by a vector, where
  each component in the vector is the frequency of
  a unique term within that document.
• Physically similar documents are plotted close to
  each other given certain parameters and
  consequently trends are discovered within a text
  collection that are difficult to discover via
  traditional means.

15 Million Documents
28
Melanoma Research

• RENCI Collaboration with Lineberger Comprehensive
  Cancer Center
• Develop 3D Visualization Tools to study DNA
  damage associated with Melanoma based on
  integrated genomic data
• Increase our understanding of skin cancer
• Lead to new strategies for protecting humans from
  genotoxins that trigger a DNA damage response in
  cells

29
Soundings Project

• Collaboration with National Humanities Center,
  UNC Libraries
• Digitize entire audio archive of Soundings
  national radio show
• multiple source formats LP, 7 tape, CD-ROM,
  paper
• Create and publish free web database
• digital audio - multiple formats, e.g. WAV, MP3
• metadata - generated from source materials
• transcriptions - automatically generated text
  with DocSoft system

30
The Data Tsunami

• Many sources
• business
• agricultural
• biomedical
• environmental
• engineering
• manufacturing
• financial
• social and policy
• historical
• Many causes and enablers
• increased detector resolution
• increased storage capability
• The challenge extracting insight!

31
Technology curves
0
1
2
3
4
5
Number of Years
32
Some data points

• Moores Law
• Capability doubles every 18 months
• Gilders Law
• communications power doubles every six months
• Metcalfes Law
• the usefulness, or utility, of a network equals
  the square of the number of users
• Reeds Law
• Utility of large networks, particularly social
  networks, can scale exponentially with the size
  of the network
• Storage Law
• Density doubles every 12 months, costs fall by
  10X each year

33
Data Storage

• Prediction The cost for 128 kilobytes of memory
  will fall below 100 in the near future.
• Creative Computing magazine December 1981, page 6

Cost per GigaByte
34
Digital Reality The Exponentials
1956
1972
2006

• Megabyte
• a small novel
• Gigabyte
• a pickup truck filled with paper or a DVD
• Terabyte one thousand gigabytes 500 today
• entire U.S. Library of Congress is ten terabytes
  of text
• 260,000 songs 1 TB
• Petabyte one thousand terabytes
• 1-2 petabytes equals all academic research
  library holdings
• Exabyte one thousand petabytes
• 5 exabytes of words spoken in the history of
  humanity

Source Hal Varian, UC-Berkeley
35
Digital Factoids (1)

• In 2007 the amount of information created,
  captured or replicated exceeded capacity
• 281 Exabytes of information created in 2007
• 45 GB per person across the globe
• Diversity of commercial data and file sizes is
  enormous
• Movies (GBs), photos (MBs), email (KBs), RFID
  (bps)
• Number of file packages and types is rapidly
  growing
• Estimated that 1.8 Zettabtyes will be created in
  2011
• 6X from 2007 (Billion Gigabytes)
• almost half the data generated in 2011 will not
  have a permanent home

The Diverse and Exploding Digital Universe IDC
Paper
36
Digital Factoids (2)

• TeraBytes
• Photos updated to Facebook 20
• each month
• All the videos on YouTube 530
• Data collected by the Hubble 120
• Data generated each week by the 300
• Large Hadron Collider (LHC)
• Data processed by Google
• every 72 min 1,000
• Power consumption for data centers per rack was 1
  KW in 2000 it is now 10 KW per rack with plans
  to move to 20 KW a major data center requires as
  much power as an aluminum smelting plant!

37
Sensor Data Overload
Source Chris Johnson, Utah Art
Toga, UCLA
Source Robert Morris, IBM

• High resolution brain imaging
• 4.5 petabytes (PB) per brain

38
Biological Models

• Simulation and prediction
• structures and dynamics
• Reasoning and discovery
• reverse engineering

Temporal (seconds)
Spatial (nM3)
39
History of Social Networks
Webkinz
Linkedin
Club Penguin
Friendster
My Space
Live Journal
Twitter
Ning
AOL IM
The Well
Facebook
ICQ
First Wiki
Usenet
1980 1985 1990 1995 2000 2005 2007
40
Some observations (1)

• Ownership of data plus low cost fuels growth and
  number of data systems
• Ownership of data will continue to drive
  distribution and local control
• More people will want to be able to access more
  data
• Databases and data collections will more
  aggressively address federation and
  interoperability
• Centralized systems reaching scaling issues while
  end terminal devices continue to increase power
• Multi-disciplinary teams become more mainstream
  for leading edge research
• Mobility and personal control will continue to
  drive innovation and business

41
Some Observations (2)

• Every intelligent person can amass the sort of
  information that used to require travel,
  networking and research assistants
• Competitiveness and success will come to those
  who
• Can put together the best team
• Can marshal the best resources and capabilities
• Genomic discovery wont be a research problem, it
  will be a search function
• Data anonymity and private information may become
  a relic of the past
• What does it mean to be a member of a community
  when the community is virtual and global?