NUMA obliviousness through memory mapping

With the rise of multi-socket multi-core CPUs a lot of ef-fort is being put into how to best exploit their abundant CPU power. In a shared memory setting the multi-socket CPUs are equipped with their own memory module, and ac-cess memory modules across sockets in a non-uniform access pattern (NUMA). Memory access across socket is relatively expensive compared to memory access within a socket. One of the common solutions to minimize across socket memory access is to partition the data, such that the data affinity is maintained per socket. In this paper we explore the role of memory mapped stor-age to provide transparent data access in a NUMA environ-ment, without the need of explicit data partitioning. We compare the performance of a database engine in a dis-tributed setting in a multi-socket environment, with a database engine in a NUMA oblivious setting. We show that though the operating system tries to keep the data affinity to local sockets, a significant remote memory access still occurs, as the number of threads increase. Hence, setting explicit pro-cess and memory affinity results into a robust execution in NUMA oblivious plans. We use micro-experiments and SQL queries from the TPC-H benchmark to provide an in-depth experimental exploration of the landscape, in a four socket Intel machine