Information & Knowledge Technologies Business Unit Distributed Systems Technology Group Projects

Adaptive Quality of Service for Availability (AQuA)

BBN Distributed Systems Project -- Technical Overview

Current large, C3 distributed applications can not specify the availability they require from remote objects and subsystems, and cannot adapt to changes in resource availability. The objective of this project is to make orders of magnitute improvements in the specification of availability requirements, in the prediction of availability under varying conditions, and in the adaptability of the applications and resource management strategies by providing infrastructure mechanisms to support dynamic behavior. The goal is to develop a will be used to extend the CORBA based Application Program Interface (API) for specifying an application's availbility requirements through defining a set of acceptable operating regions, and adapting when various thresholds of predicted availability will not be met.

Our approach is based on two key ideas: start from a modern, flexible and evolving commercially available distributed object computing substrate (OMG CORBA), and base the new infrastructure around the integration of concepts previously developed in three complementary research projects:

• Quality Objects (QuO) architecture, from BBN's Distributed Systems Department which provides a framework for quality of service oriented object resource management,
• Horus, a communications substrate from Cornell University's Department of Computer Science, which provides group and object replication management services, and
• UltraSAN, a toolkit from University of Illinois at Urbana-Champaign's Center for Reliable and High-Performance Computing, which provides a performability modeling environment for predictive modeling.

QuO will be enhanced to support adaptable availability requirements. Horus will be enhanced to support operation over Wide Area Communication. UltraSAN will be enhanced to allow the use of hierarchical, multi-module, model specifications and to predict the availability of specific system configurations. All three will be integrated together in a concept demonstration. A configurable testbed with adjustable availability characteristics and fault injection capabilities is being established to mutually calibrate and validate the models, and to demonstrate the software.

Background Papers

• John A. Zinky, David E. Bakken, and Richard Schantz.
  Architectural Support for Quality of Service for CORBA Objects
  Theory and Practice of Object Systems, To appear April, 1997.
• Quad Chart

AQuA People

BBN

• Rick Schantz
• Dave Bakken
• David Karr

University of Illinois

• Bill Sanders
• Michel Cukier
• Dan Deavours
• Harpreet Duggal

Cornell University

• Ken Birman
• Alexey Vaysburd

This project is a DARPA/ITO-funded research effort.

© 2010 BBN Technologies