Towards designing dependable software systems

虽然现代计算机系统已经与我们的日常生活紧密地交织在一起，但它们是不安全的，容易发生故障，难以配置，因此很难可靠。本研究的目的是提高软件系统的可靠性。我们将专注于设计和构建能够分析、检测和从计算机入侵、软件错误配置和软件错误中恢复的系统。传统的解决办法侧重于防止这些可靠性问题的发生，但不幸的是，这些问题没有决定性的解决办法，而是必须定期解决的持续的生活事实。例如，一个软件系统可以不断地加强，但这并不能消除入侵的可能性。相反，允许详细的事后分析和恢复的系统可以更灵活地应对软件错误。

While modern computer systems have become deeply intertwined with our daily lives, they are insecure, failure-prone and hard to configure and thus hardly dependable. The goal of this research is to improve dependability in software systems. We will focus on designing and building systems that allow analysis, detection and recovery from computer intrusions, software misconfiguration and software bugs. Traditional solutions have focused preventing the occurrence of these dependability problems, but unfortunately, these problems have no decisive solutions but are ongoing facts of life that must be resolved regularly. For example, a software system can be hardened endlessly but that does not eliminate the possibility of an intrusion. Instead, systems that allow detailed post-facto analysis and recovery can be made more resilient to software errors.