C.: Live migration of virtual machine based on full system trace and replay

DMCA

C.: Live migration of virtual machine based on full system trace and replay (2009)

Cached

by Hai Jin , Haikun Liu , Xiaofei Liao , Liting Hu , Peng Li

Venue:	In: Proceedings of the 18th ACM international
Citations:	36 - 2 self

BibTeX

@INPROCEEDINGS{Jin09c.:live,
    author = {Hai Jin and Haikun Liu and Xiaofei Liao and Liting Hu and Peng Li},
    title = {C.: Live migration of virtual machine based on full system trace and replay},
    booktitle = {In: Proceedings of the 18th ACM international},
    year = {2009},
    pages = {101--110}
}

OpenURL

Abstract

Live migration of virtual machines (VM) across distinct physical hosts provides a significant new benefit for administrators of data centers and clusters. However, previous memory-to-memory ap-proaches only make the live VM migration be applied in local area networks (LAN), and their im-proved methods for wide area networks (WAN) environment bring the VM migration with a long period of downtime. This paper describes the design and implementation of a novel approach that adopts checkpointing/recovery and trace/replay technology to provide fast, transparent VM migra-tion for both LAN and WAN environments. With execution trace logged on the source host, a syn-chronization algorithm is performed to orchestrate the running source and target VM until they get a consistent state. This scheme can greatly reduce the migration downtime and network bandwidth consumption. Theoretic analysis indicates that the service downtime to migrate a VM across WANs is equivalent to the downtime cost while migrating a VM in a LAN. Experimental measurements show that for a variety of workloads migrated in a LAN or across WANs, the migration downtime is only 200 milliseconds with a little difference. 1.

Keyphrases

virtual machine live migration full system trace migration downtime experimental measurement previous memory-to-memory ap-proaches source host wan environment vm migration service downtime novel approach im-proved method execution trace data center long period distinct physical host live vm migration little difference transparent vm migra-tion network bandwidth consumption consistent state significant new benefit target vm local area network theoretic analysis downtime cost running source syn-chronization algorithm trace replay technology wide area network

DMCA