TRUST, Washington, D.C. Meeting January 9–10, 2006
Air Force Center for Research on GIG/NCES Challenges (AF-TRUST-GNC)
Ken Birman
"TRUST and the Global Information Grid", Ken Birman
2TRUST, Washington, D.C. Meeting January 9–10, 2006
Context
TRUST mission includes commitment to create significant dialog with stakeholder communities (e.g. in medicine, financial community, power systems)
2005 saw substantial progress with Air Force– Dialog builds on a longer history of collaboration between Air
Force and our participants. For example, Air Force Information Assurance Institute at Cornell
– TRUST members assisted in two major studies of GIG/NCES impact on Air Force research priorities in 2005
AFRL/IF (JBI) Prometheus study Info Sharing 2010 study requested by SAF-XCX: a pair of TLAs
that includes CIO office of the Secretary of the Air Force (SAF)). AF-TRUST proposal reflects priorities identified in
these studies
"TRUST and the Global Information Grid", Ken Birman
3TRUST, Washington, D.C. Meeting January 9–10, 2006
Advantage: Information
“The natural formation of the country is the soldier's best ally; but a power of estimating the adversary, of controlling the forces of victory, and of shrewdly calculating difficulties, dangers and distances, constitutes the test of a great general. He who knows these things, and in fighting puts his knowledge into practice, will win his battles. ”
- General Sun-Tzu Wu, 512BC
Challenge? Finding it!
"TRUST and the Global Information Grid", Ken Birman
4TRUST, Washington, D.C. Meeting January 9–10, 2006
The proposed NCES/GIG architecture?
Basis is Web Services standard, although CORBA is likely to be used on server clusters
Primary application platform will be Microsoft Windows
NSA and DISA are playing key roles in mapping these components to military needs
"TRUST and the Global Information Grid", Ken Birman
5TRUST, Washington, D.C. Meeting January 9–10, 2006
Steps to GIG/NCES
Today– AF runs three side-by-side operational networks and many
dedicated subnetworks, i.e. to control autonomous vehicles– Connects to Internet and other services through various
DISA-operated gateways. Tomorrow
– Single Dark Core: A vision of a unified network with a small high-security core and a substantial audited but medium-security region.
– XML browsers and email throughout, posing a recognized security risk but offering needed information accessibility
– Legacy/stovepipe applications ported and wrapped for accessibility within this common operating environment
"TRUST and the Global Information Grid", Ken Birman
6TRUST, Washington, D.C. Meeting January 9–10, 2006
Missing pieces to the story?
There are many– At a “superficial” level, just using the proposed
platform to solve the kinds of problems being posed is challenging
For example, imagine an application that needs mapping data for Falluja. Which servers have this data? Are some more up to date, or less loaded, or experiencing faults? Which one is best? What security policies should apply?
– At a more technical level, Web Services lack properties one would normally expect for mission-critical military systems
TRUST, Washington, D.C. Meeting January 9–10, 2006
The Prometheus Project
Tuesday, January 10, 2006
Ken [email protected] University
Mike [email protected] Mellon University
Douglas C. [email protected] University
Real-time, Scalable, & Secure Information
Management for the GIG
"TRUST and the Global Information Grid", Ken Birman
8TRUST, Washington, D.C. Meeting January 9–10, 2006
Prometheus Emphasis: Meeting Demands for QoS-enabled Information Management
Key solution space challenges• Enormous accidental & inherent
complexities• Continuous technology evolution &
change• Highly diverse network, platform,
language, & tool environments
Key problem space challenges• Network-centric, dynamic, very large-
scale systems of systems (SoS)• Stringent simultaneous quality of
service (QoS) demands• e.g. real-time, scalability, security
• Demand for QoS-enabled operational & tactical Global Information Grid (GIG)
"TRUST and the Global Information Grid", Ken Birman
9TRUST, Washington, D.C. Meeting January 9–10, 2006
Results include a mix of prototypes, experiments, & analyses
Summary of Results from Prometheus Study
We decided to drill down on four primary questions
QoS-enabled Publish/Subscribe Technologies for Tactical Information Management
Scalable Fault- & Intrusion-Tolerance for Critical GIG Services
Scalable Enterprise Service-Oriented Architectures
Investigating a Unified Framework for Demonstrating Policy Compliance
"TRUST and the Global Information Grid", Ken Birman
10TRUST, Washington, D.C. Meeting January 9–10, 2006
Coordination Of Multiple UAVs
Dynamic MissionReplanning
Feedback &Control
Image Processing & Tracking
Focus Area 1: QoS-enabled Publish/Subscribe Technologies for Tactical Information Management
DARPA PCES Capstone demo, April 14, ‘05, White Sands Missile Range
"TRUST and the Global Information Grid", Ken Birman
11TRUST, Washington, D.C. Meeting January 9–10, 2006
Focus Area 2: Scalable Fault- & Intrusion- Tolerance for Critical GIG Services
Critical GIG services must survive failures & attacks
An intrusion-tolerant service is one that continues to operate correctly despite the corruption of some of its components
– “Intrusions” modeled as Byzantine faults (arbitrary behavior)
In this focus area, we have studied how to build fault- & intrusion-tolerant services to be fault-scalable
– i.e., service performance scales as number of faults tolerated grows
Application
ApplicationApplication
Application
Application
Pub/Sub Service
Application
Application
Application
"TRUST and the Global Information Grid", Ken Birman
12TRUST, Washington, D.C. Meeting January 9–10, 2006
Tactical Network
Satellite Network
Terrestrial IP Network
TerrestrialCircuit Network
Red LAN
Access FunctionsAccess Functions
Red LAN
Access FunctionsAccess Functions
Protection of Data-in-TransitProtection of
Data-in-Transit
IA Policy-based Routing
IA Policy-based Routing
COI level Connectivity, Bandwidth, Priority
Enforcement
COI level Connectivity, Bandwidth, Priority
Enforcement
SERVICE
Service Allocation & Prioritization
Service Allocation & Prioritization
GIG
Goals Enhance SOA platforms to support policy-driven enforcement of access to GIG
resources integrated across information, service, & transport Enable ability to dynamically adjust SOA resource allocation mechanisms in
response to changing mission priorities, failures, attacks, etc.
Focus Area 3: Scalable Enterprise Service-Oriented Architectures
"TRUST and the Global Information Grid", Ken Birman
13TRUST, Washington, D.C. Meeting January 9–10, 2006
AF-TRUST-GNC
Proposal focuses on three areas, roughly corresponding to the ones identified by Prometheus– Develop algorithms and software for scalable, real-
time and fault-tolerance QoS– Investigate issues associated with very large scale
information assurance and security policy management
– Develop new technologies for scalable and secure discovery, information architectures and mediation
"TRUST and the Global Information Grid", Ken Birman
14TRUST, Washington, D.C. Meeting January 9–10, 2006
Example questions
Can we bring build a new generation of time-critical web service cluster solutions that combine time-critical update algorithms with mechanisms for handling faults and load surges?
Suppose that a vast collection of firewalls and audit mechanisms are controlled from an enormous distributed database of policies. How can we administer and update the policy databases without accidental error?
Is it possible to somehow isolate legacy applications while still enjoying the benefits of universal connectivity and access available in Web Services?
"TRUST and the Global Information Grid", Ken Birman
15TRUST, Washington, D.C. Meeting January 9–10, 2006
Proposal?
We are proposing to create a TRUST Center focused on the needs of Air Force and other military vendors as the GIG/NCES rollout occurs
We have the breadth of talent and resources to make this work and can exploit a “dream team” that unites the top research groups in the country and focuses them on AFRL priorities. AF-TRUST-GNC includes some TRUST researchers but also some new faces
We suggest a structure parallel to that used by NSF in the basic TRUST framework
"TRUST and the Global Information Grid", Ken Birman
16TRUST, Washington, D.C. Meeting January 9–10, 2006
Possible goal for 2006?
One option is to explore a similar structure with Dept. of Treasury– Very likely to build on their eCavern project– Wide range of very exciting issues in areas such
as identity-theft, data mining to enforce risk-management, regulatory and security policies, data replication and associated policy enforcement
– Could exploit facilities right on Wall Street, where both Cornell and CMU have offices in shared bldg