Reliable Computer Systems Laboratory

We're seeking project students for Fall 2009 or Spring 2010. You can find the list of available research projects here.

Despite increasing reliance on today's computing platforms, large software systems remain unreliable, and will continue to do so in the foreseeable future. Unreliable software has been reported to take lives and cost billions of dollars annually. Reliability is one of the most important problems in Computer Science.

Our group is interested in developing new technology in operating systems and software infrastructures, with particular emphasis on high availability, reliability, security, and manageability. One of our primary focus has been to develop automated techniques to effectively detect serious software bugs. In doing so, we combine fundamental systems insights with pragmatic adaption of programming analysis techniques, and have developed several novel, practical systems to improve software reliability. These systems have found numerous severe defects in real-world software in many domains, including security vulnerabilities in operating systems with millions of users, data-loss errors in a broad range of widely used storage systems, and communication protocol errors in distributed systems that manage production Windows Live clusters with more than 100K machines. Many of these defects led to immediate patches because developers were concerned about their potential damage. As a result, the software that people use daily becomes more reliable and secure.

• [Funding] July 09: The Guanyin proposal led by Junfeng with two other professors (Prof. Gail Kaiser and Prof. Jason Nieh) received $1,012K fund from NSF.
• [Paper] Dec 08: MODIST got into NSDI 09.
• [Software] Nov 08: eXplode is released.