Introduction

Multi-Agent Grid Engine library is an experiment aimed to make a programming tool for creation of autonomic systems. We define autonomic system as the system that has the following features:

• The system is composed from a set of intellectual agents. All decision-making in the system is distributed and has a form of self-organization.
• The system is able to adapt to the surrounding environment. This includes adaptation to CPU, memory and disk load, and node failure (self-healing). The system is allowed to allocate additional nodes or redistribute current resources.

In libMAGE every agent of the system can be viewed as a living cell in a biological organism. Every agent contains enough information for construction of the whole organism, however after going through the process of growth, which mimics morphogenesis, the agent gets specialized. Specialized agents form groups and function cooperatively.

MAGE library is planned to go through the following stages of development:

• 0.1.0 — At first, preliminary stage, it is planned to provide portable multi-threading support.
• 0.2.0 — I/O Automaton abstraction implementation should lay the foundation of the second release.
• 0.3.0 — Message-passing channel automata together with architecture-independent network data transfer support, TCP and MT(multi-threaded) based, would be implemented at the third development stage.
• 0.4.0 — At the next (fourth) stage IOA library would be extended by adding transmission-control and routing automata.
• 0.5.0 — Dynamic group communication should be implemented at the fifth development stage. This will allow IOA to form dynamic groups and adapt to node addition and removal.
• 0.6.0 — The sixth stage should deal with seamless system evolution. The system should be able to allocate/free resources dynamically, adapting to the changing environment.
• 0.7.0 — The final development stage should provide the library with different group templates. These will include data and code replication and fragmentation groups.