hyperCache: A Hypervisor-Level Virtualized I/O Cache on KVM/QEMU

While VDI offers several benefits, such as the increased resource utilization and the private data protection, there exist problems that can deteriorate the system performance, including I/O virtualization overhead. Before I/O requests issued in VMs are completed in VDI, they should go through multi...

Full description

Saved in:

Bibliographic Details
Published in:	2018 Tenth International Conference on Ubiquitous and Future Networks (ICUFN) pp. 846 - 850
Main Authors:	Taehoon Kim, Seungho Choi, Jaechun No, Sung-Soon Park
Format:	Conference Proceeding
Language:	English
Published:	IEEE 01-07-2018
Subjects:	access frequency Bandwidth Emulation Kernel Metadata metadata repository Performance evaluation QEMU Virtual machine monitors Virtualization writeback
Online Access:	Get full text
Tags:	Add Tag No Tags, Be the first to tag this record!

Description
Summary:	While VDI offers several benefits, such as the increased resource utilization and the private data protection, there exist problems that can deteriorate the system performance, including I/O virtualization overhead. Before I/O requests issued in VMs are completed in VDI, they should go through multiple software layers, such as the layer from the backend, shared storage to the host server and the layers between guest operating system, hypervisor, and eventually host operating system. In this paper, we present a hypervisor-level cache, called hyperCache, which is possible to provide a shortcut in KVM/QEMU by intercepting I/O requests in the hypervisor, while taking into account the I/O access frequency. Our experimental results demonstrate that our design improves I/O bandwidth over the existing QEMU.
ISSN:	2165-8536
DOI:	10.1109/ICUFN.2018.8436697