Measuring and Improving Memory's Resistance to Software Corruption; Relational Instruction

系统设计人员长期以来一直认为，由于内存的易失性和软件损坏的脆弱性，内存无法可靠地存储永久数据。 内存的不可靠性通过迫使某些应用程序将数据写入磁盘来降低系统性能，并通过迫使关心性能的应用程序在崩溃时丢弃30秒的数据来损害可靠性。本职业计划的主要研究目标是：1）即使在存在操作系统或数据库错误的情况下，内存也能可靠地存储永久数据; 2）测量软件故障损坏内存的频率，包括有无额外保护。 本职业生涯计划的主要教育技术是关系教学： 花时间和精力与学生个人联系。 该计划提出了几种简单的方法，如集体午餐，以有效地做到这一点与大型，本科班;以前的经验证明，如何照顾学生在个人的方式激励他们学习。

System designers have long assumed that memory cannot reliably store permanent data because of memory's volatility and vulnerability to software corruption. Memory's unreliability slows system performance by forcing some applications to write data through to disk and hurts reliability for forcing applications that care about performance to throw away 30 seconds of data on a crash. The primary research goals of this CAREER plan are 1) to enable memory to reliably store permanent data even in the presence of operating system or database errors; and 2) to measure how often software faults corrupt memory, both with and without extra protection. The primary education technique of this CAREER plan is relation instruction: spending time and effort relating with students personally. This plan suggests several simple ways, such as group lunches, to do this effectively with large, undergraduate classes; prior experience attests to how caring for students in a personal manner motivates them to learn.