Towards a Single-Level Database Architecture on Non-Volatile Memory




Abstract: The rise of Big Data emphasized the importance of memory-centric systems with an ever increasing need for more main memory, pushing DRAM to its scalability limits. New NVM technologies are emerging as an alternative to extend these limits. They combine the non-volatility and density of traditional storage with the low latency and byte-addressability of DRAM. In this presentation we show how to leverage NVM in databases as a universal memory, and make the case for hybrid NVM-DRAM data structures.

Bio: Ismail Oukid is a PhD candidate at TU Dresden, working in close collaboration with SAP and Intel on leveraging future non-volatile memory technologies as universal memory in main-memory database systems. His areas of interest include software optimization for modern hardware. He received a Master's degree from Grenoble INP - Ensimag in 2013.