Emerging trends in computer design and use are likely to make exceptions, once rare, the norm, especially as the system size grows. Due to exceptions, arising from hardware faults, approximate computing, dynamic resource management, etc., successful and error-free
execution of programs may no longer be assured. Yet, designers will want to tolerate the exceptions so that the programs execute completely, efficiently and without external intervention.
In this talk, the noted humanist James Mickens will provide core insights into problems that have plagued civilization for millennia. Using popular buzzwords like “the cloud,” “mobile,” “social,” and “mobile social clouds,” Mickens will explain why his research is fundamental to any scientific or artistic endeavor that anyone has ever attempted or will attempt in the future. Mickens will discuss his work on datacenter-scale storage systems, and why this work will likely result in a Nobel Prize.
The next-generation non-volatile memory (NVM) promises performance close to DRAM, persistence, and high density. However, providing the level of reliability and availability that large-scale storage system demands is challenging. In this talk, I will introduce Mojim, a system that preserves the performance of NVMs while providing the reliability and availability that modern storage systems require. Mojim achieves good performance by decoupling reliability, availability, redundancy, and consistency, and prioritizing reliability and availability over consistency.