U.S. Department of the Interior

1
A Cognitive Agent Based Geospatial Data
Distribution System 12 May 2006
U.S. Department of the Interior U.S. Geological
Survey
2
Geospatial Infrastructure - A possible vision of
the future...
The National will have an infrastructure that is
an information society that stimulates the
economy, improves environmental health, and
increases the general Welfare of the Nation by
communicating geographic science and enabling
problem solving within its communities.
3
It's All in the GOALS or..I'm not a

Cartographer Geographer Computer Scientist
It is in fact in choosing the goal of what to
design which is where experts make their
mistakes. Amateur designers will get the
details wrong because they just don't have the
professional mastery. Experts will typically get
the details right, put a lot of polish on the
design and get the whole project wrong.

• Here is where experts go wrong
• Miss fresh vision e.g., microcomputer
  revolution
• Vision not high enough e.g., OS/360 JCL

from The Design of Design, Frederick P. Brooks,
1999 ACM Turing Award Lecture
4
The Design Problem

• The infrastructure is a
• logistics problem!
• How to
• allocate and marshal resources
• recover from component failure
• monitor performance
• manage intellectual property rights
• control the system

5
Mgt. Governance
Elevation Data
Geospatial Infrastructure
Orthoimagery
Homeland Security
Problem Solving Communities
Core members providing governance and persistence
Provider Communities
Road Data
Education
Law Enforcement
5
6
A Business Partner utilizing The infrastructure
DHS
Network Operations Center
NGA
Support Services
PKI
Archive Services
Core Members
Commercial Archive
Deep Archives
7
Mgt. Governance
Geospatial Infrastructure
Communities
Affiliates
USGS as an affiliate
OK!
X
Not Allowed
Federal User
Special Access Arrangements
User
Core
Access Restrictions
Firewall
FISMA and The Business Model
7
8
The Infrastructure Should NOT Be

• A geographic information system
• A decision support system
• A library of links to other's data
• An information discovery system
• Web mapping application
• Monolithic

9
The Infrastructure Should Be

• An information society
• Marketplace
• A framework for delivery of products and services
• A framework for archiving quality-controlled
  spatial information
• Focused on geographic problem solving
• A system of systems

10
Information Characteristics

• Durable does not change over time
• Persistent does not disappear
• Ubiquitous available everywhere needed
• Access Managed controlled IP rights (DRM)
• Authorize known users
• Authoritative Content The correct data

11
Digital Rights Management Architecture Core
Entities Model
Rights
Owns
Creator
Over
User
Usage Type
Content
After Lannella, Renato Digital Rights Management
(DRM) Architectures, D-Lib Magazine, June 2001
12
System Requirements for The Infrastructure

• Geographic independence for availability,
  durability, partnerships, and resistance to
  failure.
• Encryption for privacy, signatures for
  authenticity, and Digital Rights Management.
• Redundancy with continuous repair and
  redistribution for long-term durability.
• Automatic optimization, diagnosis and repair.

13
Network Architectural Elements
13
14
Characteristics needed byThe Infrastructure

• Partners bring multiple organizational management
  structures
• Cant possibly manage large numbers of affiliate
  servers by hand!
• System should automatically
• Adapt to failure of individual components
• Repair itself
• Allow affiliates to join and leave without prior
  notice
• Adapt to changes in demand and regional outages

15
Characteristics needed by The Infrastructure

• Guarantee data is available for 100s of years
  (multiple generations)
• New servers added from time to time
• Old servers removed from time to time
• Affiliates can self associate (join and leave)
• Redundant components with geographic separation
• System not disabled by natural or man-made
  disasters
• Gain in stability through statistics

16
Characteristics needed by The Infrastructure

• Untrusted Infrastructure
• Comprised of untrusted components
• Only cipher text within the infrastructure
• Mostly Well-Connected
• Data producers and consumers are connected to a
  high-bandwidth network most of the time
• Exploit multicast for quicker consistency when
  possible
• Provide for unconnected operation

17
Characteristics needed by The Infrastructure

• Nomadic Data
• Promiscuous caching
• data can be cached anywhere, anytime
• Floating replicas
• object replicas independent of the server they
  reside on
• Optimize locality and to trade off consistency
  for availability
• Infrastructure should disappear into the
  background
• Dont want to worry about backup.
• Dont want to worry about obsolescence.
• Need lots of resources to make data secure and
  highly available, BUT dont want to own them.

18
Autonomic Computing Requirements for the
Infrastructure
19
An Architecture for The Infrastructure
White Pages
Yellow Pages
Agent
Agent
Affiliate
Blackboard (PLAN)
Blackboard (PLAN)
Executor
Planner
Subscribe
Publish
Publish
Subscribe
Plug-ins
Plug-ins
Message Queue
Affiliate
Message Queue
Viewer
GIS System
Data Catalogue
Core Members
Control
Core Services
20
For more information

• OceanStore vision paper for ASPLOS 2000
• OceanStore An Architecture for Global-Scale
  Persistent Storage
• OceanStore web site http//oceanstore.cs.berkele
  y.edu
• Cougaar web site
• http//www.cougaar.org
• Fedora web site
• http//www.fedora.info

21
(No Transcript)