Paper Title: Achieving Fault Tolerance for Storage Class Memory

Paper Abstract: Storage Class Memory (SCM) allows designers to replace several tiers of the memory hierarchy with a single, cost-effective, uniform memory/storage. To make such large-scale deployments practical, however, designers must be able to deal with unavoidable SCM wearout and failures. A new approach can provide fault tolerance in SCM-based main memory. It treats memory as a distributed storage system and relies on replication, plus programmable interconnect providing fast consensus to keep the replicas consistent. Initial experiments demonstrate reasonable overhead over local memory access and show great promise as scalable, fault-tolerant main memory.

Paper Author: Huynh Tu Dang, Technologist, Western Digital

Author Bio: Huynh Tu Dang is a Principal Engineer at Western Digital, where he develops network-accelerated fault-tolerant data management systems. His research interests are in dependable distributed systems and computer networking. He has pioneered a novel approach to accelerate performance of consensus systems using a new type of programmable switch. He earned a PhD in computer science from Università Della Svizzera Italiana (USI in Lugano, Switzerland), and an MS in computer science from Polytech Nice-Sophia Antipolis (Nice, France). He has published many articles in both journals and conference proceedings and has applied for two patents.