Francine Berman

1
The Cal-IT2 Software Challenge

• Francine Berman
• UCSD Interfaces and Software Layer Leader

2
Software and Cal-IT2

• Cal-IT2 will develop living laboratories which
  prototype future work, play and Internet
  environments
• Software will provide the critical glue which
• enables system components to work together
• enables applications to achieve performance
• The Cal-IT2 SW Environment Will
• Be completely decentralized
• Target
• an extraordinary diversity of HW and SW (sensors,
  agents, PDAs, servers, wireless and wireline
  networks, control rooms, )
• continuous streams of data from millions of
  sensors
• Require guerrilla infrastructure which
  self-assembles from the bottom up
• Be Use at your own risk
• Pave the way for commercial deployment

3
Cal-IT2 Applications

• Cal-IT2 Applications will push the envelope
• Applications will process data from massive
  number of throwaway endpoints
• E.g. traffic management, disaster response
• Seti_at_home, Entropia applications
• Applications will compute based on emergent
  behavior of millions of self-organizing
  components
• Think Distributed Interactive Simulation
• Applications will compute on wide variety of
  heterogeneous resources simultaneously
• PCs, PDAs, Servers, Supercomputers, remote
  instruments
• High-capacity networks, storage and computational
  resources will enable previously big iron
  applications to transition efficiently to the
  Grid
• Grid-enabled CFD, genome_at_home

4
Cal-IT2 Software Challenges

• New application paradigms
• Self-organization, emergent behavior,
  megacomputing
• Statistical approach in which some fraction of
  devices always non-functional
• New models for design, prototyping, simulation
• Security
• Authentication of people and devices, integrity
  of sensor data, distributed intrusion detection,
  
• Data
• New approaches to collecting, storing, accessing
  and analysis of huge amounts of sensor data
• Scalable Grid computing
• Development of large-scale campus-wide
  distributed testbeds
• Megacomputing Grid computing with millions of
  throw-away end-devices
• Agent-based technology
• Adaptive computing, mobile agents
• Interfaces
• Large screen control rooms for disaster
  response, ultimately a prototype for smart
  offices
• Access to real-time analysis of very large
  datasets, visualization, multi-scale information

5
The Grid An Initial Model for Cal-IT2 SW
A source book for the history of the future --
Vint Cerf
Edited by Ian Foster and Carl Kesselman www.mkp.co
m/grids

• SW focus on
• Services
• Component Integration
• De-centralization
• System and User-level Middleware

6
MegaComputing Taking the the Grid to the Extreme

• Megacomputing Applications will be
  structured
• for throwaway end devices
• for continuous execution
• to achieve performance and
• computational goals in highly
• dynamic, unfriendly envts.

Seti_at_Home -- sustains over 2 teraflops in
production use
Genome_at_home -- bringing big iron applications
into Grid environments
Entropias Great Mersenne (2p 1) Prime Search
found the firstmillion digit prime
7
The Software Challenge is immense

• First Steps
• SensornetSim Voelker
• Simulator for sensor arrays which can support the
  development of prototypes of sensor
  infrastructure and SW components for Cal-IT2
• Expansion of Sanctuary Security System to
  wireless Yee
• Extend UCSD Sanctuary secure mobile agent system
  as a wireless application security platform
• Development of applications which use mobile code
  for latency hiding
• Prototype smart campus MUDD for 6th College
  Griswold, Voelker, Wienhausen/6th, etc.
• First year centralized MUDD for 6th college
  courses, location sensing for students with PDAs
• Plan for 21st Century UCSD course at 6th College

8
The Software Challenge is immense

• First Steps
• Genomics Applications
• Development of assembler for sequencing large
  genomes Pevzner
• Detection of protein structure from sequences
  using Protein centroids Rosen/CSE, Bourne/SDSC
• Coupled Oceanographic Modeling Simulation
  Baden/CSE, Stammer/SIO
• Assimilation of satellite data into running
  oceanographic simulations
• Content-Based Web Crawling Elkan, Belew,
  Cottrell
• Webclopedia extends Internet-accessible
  information to include billions of Web pages
  through their content
• Seeks to develop a Web of Knowledge that
  functions as a multilingual encyclopedia to
  answer simple questions

9
Cal-IT2 Interfaces and SW Layer

• To fulfill the Cal-IT2 vision, we will focus on
  fundamental research in
• Middleware
• Security
• Networking
• Modeling and simulation
• Grid Computing, Megacomputing
• Mobile agents
• Database technology and Data-mining
• Over 37 faculty involved from UCSD CSE and UCI
  CIS
• We are already getting started