Transcript A Mobile Internet Powered by a Planetary Computer

A Mobile Internet Powered by a
Planetary Computer
Invited Talk
Grid on the Go Workshop
NCSA
University of Illinois at Urbana-Champaign
Urbana, IL
May 21, 2001
The Emerging Planetary Scale Grid
• Wireless Access--Anywhere, Anytime
• Broadband to the Home and Small Businesses
• Vast Increase in Internet End Points
– Embedded Processors
– Sensors and Actuators
– Information Appliances
• Highly Parallel Light Waves Through Fiber
• Emergence of a Distributed Planetary Computer
– Storage of Data Everywhere
– Scalable Computing Power
The Next Wave of the Internet Will
Extend IP Throughout the Physical World
This is the Research Context for the
California Institute for Telecommunications
and Information Technology
Materials and Devices Team, UCSD
UC San Diego and UC Irvine
California Institute for Telecommunications and Information Technology
• 220 Faculty and Senior Researchers
• Layered Structure
–
–
–
–
–
Materials and Devices
Networked Infrastructure
Interfaces and Software
Strategic Applications
Policy
• New Funding Model (4 Years)
–
–
–
–
–
–
State $100M
Industry $140M
Private $30 M
Campus $30M
Federal $100-200M
Total $400-500M
• One of Four Awarded
The Era of Guerilla Infrastructure
• Guerilla vs. Commercial Infrastructure
–
–
–
–
–
Bottom Up
Completely Decentralized
Self-Assembling
Use at Your Own Risk
Paves the Way for Commercial Deployment
• Examples
–
–
–
–
–
NSFnetInternet
NCSA MosaicWeb
NapsterPeer-to-Peer Storage
SETI@homePeer-to-Peer Computing
IEEE 802.11Broadband Wireless Internet
Broadband Wireless Internet is Here Today
• Create Wireless Internet “Watering Holes”
– Ad Hoc IEEE 802.11 Domains
– Real Broadband--11 mbps Going to 54 mbps
– Security and Authentication can be Added
– But, it is Shared and Local
– Home, Neighborhoods, Office, Schools?
– MobileStar--Admiral Clubs, Starbucks, Major
Hotels, Restaurants, …
– UCSD—Campus Buildings, Dorms, Coffee
Shops…
The High Performance
Wireless Research and Education Network
• Cal-(IT)2 Will Build
on This Pioneering
Experiment
• Add New Science
Sensor Arrays
• Instrument Civil
Infrastructure
• Try Out New
Wireless
Technologies
• Data Analysis
NSF Funded
PI, Hans-Werner Braun, SDSC
Co-PI, Frank Vernon, SIO
45mbps Duplex Backbone
• Outreach and
Education
Roadmap to
3rd Generation Wireless (3G)
• Huge Capital Investments Already Made (Particularly in Europe)
• More Investments Required for Spectrum
• Differential Roll-Out Around the World
High Data Rate (HDR)
A 2.5G Bridge to the Future
• Qualcomm’s High Data Rate (HDR)
– Peak is 2.4 Mbps downstream, 307 kbps Upstream
– Average is 600 kbps upstream, 220 kbps down
– Extends CDMA Cellular/PCS Voice to IP Packet Data
– Can Share Existing CDMA Deployed Infrastructure
– Can be Installed in Current Cell Phones, Laptops, etc.
• CDMA2000, High Rate Packet Data Air Interface Spec.
– Telecommunications Industry Assoc. Spec. TIA/EIA/IS-856
– Also known as 1xEV
– Based on HDR
• UCSD/CalIT2 Has HDR Antennas Deployed & Working
– Testbed for Wide Area Broadband Wireless
– Use as WAN to 802.11 LAN
HDR Provides an Early View of
Broadband Wireless Internet
New Software Environments for
Wireless Application Development
• Binary Runtime Environment for Wireless (BREW)
– Works on Qualcomm CDMA Chipsets
– Middleware Between
– the Application and the Chip System Source Code
– Windows-based Software Development Kit (SDK)
– Native C/C++ applications will run most efficiently,
– Supports Integration of Java™ Applications
– Different Model of Security from JAVA
www.qualcomm.com/brew/
Will The Planned Global Rollout of 3G
Proceed as Planned?
• The Economics of Telecom
– The Huge Debt Load
– The Investment in 3G Buildout
– Is There a Business Case to Recoup?
• Technological Breakouts
– IEEE 802.11 Buildout
– Will It Skim the Cream off 3G?
– 2.5G Can Deploy Now (Sprint PCS)
– Will 3G Standardize in Europe, Asia, US?
Wireless Technologies Are
a Strong Academic Research Discipline
Center for
Wireless Communications
Two Dozen ECE and CSE Faculty
LOW-POWERED
CIRCUITRY
RF
Mixed A/D
ASIC
Materials
ANTENNAS AND
PROPAGATION
COMMUNICATION
THEORY
COMMUNICATION
NETWORKS
MULTIMEDIA
APPLICATIONS
Architecture
Changing
Modulation
Media Access
Smart Antennas
Environment
Channel
Coding
Scheduling
Adaptive Arrays
Protocols
Multiple Access End-to-End QoS Multi-Resolution
Compression
Hand-Off
Source: UCSD CWC
Creating Tiny and Inexpensive
Wireless Internet Sensors Combining…
Fluids
Stresses and Strains
0.1 mm
Optics and Lasers
UCI Integrated Nanosystems Research Facility
Integrating MEMS Sensors
With Computing, Storage, & Communication
Wireless
RTOS
Applications
sensors Reconf.
Logic
Processors
Memory
Protocol
Processors
Protocol
Stacks
Network
Transport
Data Link
Physical
RF
Protocols
SW/HW/Sensor/RF
Co-design
Reconfiguration
DSP
SoC Design
Methodologies
SW/Silicon/MEMS
Implementation
Source: Sujit Dey, UCSD ECE
Internet
As Our Bodies Move On-Line
Bioengineering and Bioinformatics Merge
• New Sensors—Israeli Video Pill
– Battery, Light, & Video Camera
– Images Stored on Hip Device
• Next Step—Putting You On-Line!
– Wireless Internet Transmission
– Key Metabolic and Physical Variables
– Model -- Dozens of 25 Processors and 60
Sensors / Actuators Inside of our Cars
www.givenimaging.com
www.bodymedia.com
• Post-Genomic Individualized Medicine
– Combine
– Genetic Code
– Body Data Flow
– Use Powerful AI Data Mining Techniques
Wireless Sensors Will Allow Instrumentation
of Critical Civil Infrastructure
New Bay Bridge Tower
with Lateral Shear Links
Cal-(IT)2 Will
Develop and Install
Wireless Sensor Arrays
Linked to
Crisis Management
Control Rooms
Source: UCSD Structural Engineering Dept.
The Perfect Storm:
Convergence of Engineering with BioMed, Physics, & IT
500x
Magnification
Nanogen MicroArray
2 mm
VCSELaser
400x
Magnification
IBM Quantum Corral
Iron Atoms on Copper
Human Rhinovirus
5 nanometers
Requires New Clean Room Facilities
Nanotechnology Will be
Essential for Photonics
VCSEL + Near-field polarizer :
Efficient polarization control,mode
stabilization, and heat management
Near-field coupling between pixels
in Form-birefringent CGH (FBCGH)
FBCGH possesses
dual-functionality
such as focusing
and beam steering
0.8
0.6
1.0
Reflectivity
TM 0th order efficiency
1.0
Near-field coupling
0.4
TE
TM
0.8
0.6
0.4
0.2
RCWA
Transparency Theory
0.0
0.2
1.3
0.60
0.65
0.70
0.75
Thickness ( mm)
0.80
1.5
1.7
1.9
2.1
Wavelength m
( m)
2.3
2.5
Micro polarizer
VCSEL
Information I/O through
surface wave, guided
wave,and optical fiber
from near-field edge and
surface coupling
FBCGH
Grating coupler
Fiber tip
Near-field
E-O coupler
+V
TM Efficiency
1.0
0.8
-V
Near-field E-O Modulator
+ micro-cavity
0.6
0.4
0.2
0.0
20
30
Angle (degree)
40
Composite nonlinear,
E-O, and artificial dielectric
materials control and
enhance near-field coupling
Source: Shaya Fainman, UCSD
Near-field E-O
modulator controls
optical properties
and near-field
micro-cavity
enhances the effect
Why the Grid is the Future
Scientific American, January 2001
The UCSD “Living Grid Laboratory”—
Fiber, Wireless, Compute, Data, Software
•Commodity Internet, Internet2
•CENIC’s ONI, Cal-REN2, Dig. Cal.
•PACI Distributed Terascale Facility
Wireless WAN
SDSC
• High-speed optical core
CS
Eng. / Cal-(IT)2
Hosp
Med
Chem
• Wireless LANs
SIO
½ Mile
Source: Phil Papadopoulos, SDSC
Near Term Goal:
Build an International Lambda Grid
• Establish PACI High Performance Network
– SDSC to NCSA LambdaNet for DTF
• Link to:
– State Dark Fiber
– Metropolitan Optical Switched Networks
– Campus Optical Grids
– International Optical Research Networks
• NSF Fund Missing Dark Fiber Links For:
– Scientific Applications
– Network Research
Optically Linked High Resolution Data
Analysis and Crisis Management Facilities
• Large-Scale Immersive Displays SDSC
– Panoram Technology
• Fiber Links Between SIO, SDSC,
SDSU
– Cox Communication
• Optical Switching
– TeraBurst Networks
• Driven by Data-Intensive
Applications
– Seismic and Civil Infrastructure
– Water Environmental System
• Integrate Access Grid for
Collaboration
SIO
Attack of the Killer Micros
From Vector SMPs to Intel Clusters
Cray X-MP
IBM SP
TMC CM-5
RISC
Processors
Time
PC Clusters
Intel
Processors
ASCI Red
Linux Clusters
Peer-to-Peer Computing and Storage
Is a Transformational Technology
The emergence of Peer-to-Peer computing
signifies a revolution in connectivity that will be
as profound to the Internet of future
as Mosaic was to the Web of the past.”
–Patrick Gelsinger, VP and CTO, Intel Corp.
Grid Computing (Condor)
For Quantum Monte Carlo Materials Codes
Input
Output
Condor
Clone 1
Clone 2
...
Clone M
• Pool of Workstations: Condor Carries
Out the Management, Distribution,
Monitoring and Checkpointing
• Very Coarse-Grain Parallelism:
Parameter Scans, Independent
Searches, Monte Carlo
• Each Clone: Independent Random
Number Streams - “Grand Averages”
Evaluated at the Very End
UW Pool ~ 800 Workstations
Www.Cs.Wisc.Edu/condor
NCSA/BI Pool ~ 40 Workstations Capable of Providing
Free 9000 SGI CPU- Hours Per Month
Torelli, Mitas, Nano Team + Livny, UW Madison
Entropia’s Planetary Computer
Grew to a Teraflop in Only Two Years
The Great Mersenne Prime (2P-1) Search (GIMPS)
Found the First Million Digit Prime
www.entropia.com
Eight 1000p IBM Blue Horizons
Deployed in Over 80 Countries
SETI@home Demonstrated that PC Internet
Computing Could Grow to Megacomputers
• Running on 500,000 PCs, ~1000 CPU Years per Day
– Over Half a Million CPU Years so far!
– 22 Teraflops sustained 24x7
• Sophisticated Data & Signal Processing Analysis
• Distributes Datasets from Arecibo Radio Telescope
Arecibo
Radio Telescope
Extending the Grid to Planetary Dimensions
Using Distributed Computing and Storage
AutoDock Application Software Has Been Downloaded to Over 20,000 PCs
Nearly 3 Million CPU-Hours Computed
In Silico
Drug Design
Art Olson,
TSRI
Monte Carlo Cellular Microphysiology
From IBM Blue Horizon to the Grid
•
PROJECT LEADERS
– Francine D. Berman
Rendered by Tom Bartol of the Salk Institute for Biological Studies &
Joel Stiles of Cornell University
using Pixar PhotoRealistic RenderMan
– UC San Diego
– Terrence J. Sejnowski
– Salk Institute for
Biological Studies
•
PARTICIPANTS
– Dorian Arnold
Jack Dongarra
Richard Wolski
– University of
Tennessee
Neurotransmitter Activity
Leading to Muscle Contraction
• MCell Simulated:
• The Transmission of 6,000 Molecules of the
Neurotransmitter Acetylcholine (Cyan Specks)
• In a Reconstructed Mouse Sternomastoid
Neuromuscular Junction
• Containing Acetylcholinesterase (White Spheres).
www.npaci.edu/envision/v16.4/mcell.html
– Thomas M. Bartol
Lin-Wei Wu
– Salk Institute for
Biological Studies
– Henri Casanova
Mark H. Ellisman
Maryann Martone
– UC San Diego
The Emerging Planetary Supercomputer
• Napster Meets SETI@Home
– Distributed Computing and Storage
• Assume Ten Million PCs in Five Years
– Average Speed Ten Gigaflop
– Average Free Storage 100 GB
• Planetary Computer Capacity
– 100 Petaflop Speed
– 1 Exabyte Storage
• Serve as Global Compute and Storage
Server for Mobile Clients