Home // CLOUD COMPUTING 2017, The Eighth International Conference on Cloud Computing, GRIDs, and Virtualization // View article

Data Placement Based on Data Semantics for NVDIMM/DRAM Hybrid Memory Architecture

Authors:
Gaku Nakagawa
Shuichi Oikawa

Keywords: memory management, nvdimm, non-volatile memory, solid state drive

Abstract:
Non-Volatile Dual Inline Memory Module(NVDIMM) makes it possible to expand the main memory with non-volatile memory. However, constructing the main memory only with NVDIMM is unrealistic because NAND Flash, the most promising candidate as NVDIMM device, has several shortcomings about write access. The hybrid memory architectures with NVDIMM and Dynamic Random Access Memory (DRAM) is a method to hide the shortcoming of NAND Flash. In the architecture, we can offload write-hot data to DRAM. In this paper, we utilize data semantics to determine data placements on NVDIMM/DRAM hybrid memory architecture. The architecture requires distributing data between NVDIMM and DRAM. Data semantics (i.e., meaning of data) is useful for the decision for the data placements. As a proof-of-concept, we executed a simulation experiment to determine data allocation between NVDIMM and DRAM based on the data semantics. As a result, we could suppress write access to the NVDIMM area under only 0.2% of the DRAM area.

Pages: 99 to 101

Copyright: Copyright (c) IARIA, 2017

Publication date: February 19, 2017

Published in: conference

ISSN: 2308-4294

ISBN: 978-1-61208-529-6

Location: Athens, Greece

Dates: from February 19, 2017 to February 23, 2017