As computing devices and networks are becoming more powerful and ubiquitous, the networking landscape is evolving into new paradigms such as autonomic networks [1], pervasive networks [2] and grid networks [3]. In these emerging paradigms, network systems will be much more complex and larger than the current ones; the author believes that the capability of network systems is going beyond the capacity of human users and administrators to configure, monitor and understand all aspects of their network systems. Therefore, future network systems need to address fundamental challenges such as autonomy - the ability to operate without human intervention; scalability - the ability to scale to a large number of network hosts and users; adaptability - the ability to adapt to dynamic changes in network conditions (e.g., resource availability and network traffic); survivability - the ability to retain operation and performance despite partial system failures (e.g., network host failures). In order to meet these challenges in future network systems, the author observes that various biological systems have already developed the mechanisms necessary to overcome those challenges [4]. Therefore, the author proposes to apply key biological principles and mechanisms for designing a new architecture for network systems. The proposed architecture, called SymbioticSphere, consists of two major system components: application service and middleware platform. Each of them is modeled as a biological entity, analogous to an individual bee in a bee colony. They are designed to follow several biological principles such as decentralization, autonomy, emergence, redundancy, natural selection and symbiosis. An application service is implemented as an autonomous software agent. Each agent implements a functional service and follows biological behaviors such as migration, replication, death, energy exchange and environment sensing. A middleware platform runs on a network host and operates agents. Each platform provides a set of runtime services that agents use to perform their services and behaviors, and implements biological behaviors such as replication, death, energy exchange and environment sensing. The objectives of this study are to design and implement SymbioticSphere and evaluate the biologically-inspired mechanisms in SymbioticSphere in terms of autonomy, scalability, adaptability and survivability. Simulation results show that agents and platforms autonomously scale to network size and demand volume and adapt to dynamic changes in the network conditions (e.g., user locations, network traffic and resource availability). They also autonomously survive partial system failures such as host failures and link failures in order to retain their availability and performance.
展开▼
