From Distributed Software Systems Group, University of Massachusetts, Boston

An Immunologically-inspired Adaptation Mechanism for Self-Regulatory and Evolvable Network Applications

Executive Summary

This project investigates and evaluates an immunologically-inspired adaptation mechanism that allows network applications to autonomously adapt to dynamic environmental changes.

Research Issues

Large-scale network applications such as data center applications and grid computing applications face several critical challenges, particularly autonomy and adaptability, as they have been increasing in complexity and scale. They are expected to autonomously adapt to dynamic environmental changes in the network (e.g., workload surges and resource extinction) in order to improve user experience, expand applications' operational longevity and reduce maintenance cost.

In order to meet this requirement (autonomous adaptability), the following questions must be answered.

• How does each network application recognize dynamic changes in the network conditions?

• How does each network application respond to a given set of environment conditions?

• How does each agent adjust its configuration to the dynamic changes of environment conditions?