Understanding and Surviving Computation in the Wild

点击翻译按钮获取中文摘要

The marriage of the personal computer and the Internet has led to explosive growth in the contacts between separately administered computing resources, creating new opportunities and risks. Applets, agents, viruses, email attachments, and downloadable software are escaping the confines of their original systems and spreading through communications networks. Computers are disabled by network-borne infections; browsers crash due to unforeseen interactions between an applet and a language implementation; application programs are broken by operating system upgrades. This computation in the wild is a far cry from the carefully isolated, controlled, and managed computer systems of the past. The connections between computer systems and living systems are superficially obvious and yet deep in their implications. This proposal argues that networked computer systems can be better understood, controlled, and developed when viewed from the perspective of living systems, treating the rich and dynamic network computer environment formed by diverse benign and malicious software collectively as a software ecosystem. Taking seriously the analogy between computer systems and living systems encourages rethinking many aspects of current computing practice ranging from operating system design to communications mechanisms to computer security. This project is aimed at developing design strategies from biology, constructing software that can survive in the wild, and developing a better understanding of the current and emerging software ecosystems. Like many researchers, we believe that the current crisis of software development is unlikely to improve significantly through incremental research using traditional methods. New approaches must be tried, and new ways of thinking about computing must be developed. Biological principles stand to unify many scattered current research efforts addressing robust operation, survivability and security, while also suggesting new avenues for research. In addition, as the size and scope of software systems continues to grow, and global computer networks continue to expand, tools and methodologies from biological research will be increasingly relevant for understanding and monitoring the results.The proposal takes the investigators' ideas and insights about living systems online, emphasizing concrete implementations that demonstrate new approaches to solving real problems. Specifically, the following projects centered on and exploring computation in the wild are proposed: Homeostasis for improved operating system survivability, Beyond network intrusion detection, Toward a living networked operating system, Software diversity for species survivability, and Tools and techniques from biology. The homeostasis project will augment a computer operating system with mechanisms similar in spirit to those of biological systems, which maintain homeostasis, by coupling system-call based process monitoring to feedback mechanisms. The second project will generalized and extend an existing prototype TCP-based intrustion-detection systems based on immunological principles. The third project differs from the first two by relaxing the constraint of staying within the bounds of operating system architectures and technology. Rather, it will pursue promising directions for robust operation and computer security based on living systems principles, even where they involve possibly significant levels of incompatibility with existing software and designs. The software diversity project will investigate methods by which software can be made more diverse, and the fifth project will apply quantitative methods from biology to the study of the current software ecosystem.