Five Years of Reliability Research: Why Storage Systems Are Broken, and What We Can Do About It

Event details
Date | 14.06.2010 |
Hour | 14:15 |
Speaker | Prof. Remzi Arpaci-Dusseau, University of Wisconsin |
Location | |
Category | Conferences - Seminars |
For over five years, we have been investigating the reliability of the software and hardware systems that store data. In almost every case, we have found serious design flaws, implementation problems, and software bugs, many of which can lead to system crashes, data loss, or silent data corruption.
In this talk, I will present an overview of our research, highlighting the aforementioned problems. I will discuss why such problems seem to repeatedly occur and thus are fundamental to the current way in which we build software. I will then describe some techniques we have eveloped to build more reliable storage systems in the future.
A sub-theme of my talk will focus on an issue of broader interest to younger researchers: where our ideas came from. Thus, in presenting the work, I will not only discuss the core technical material but also the thinking, reasoning, chance, and dumb luck that led to some of our ideas. Prof. Arpaci-Dusseau's homepage
Practical information
- General public
- Free