Research Article
Dynamic Migration of Computation through Virtualization of the Mobile Platform
@INPROCEEDINGS{10.1007/978-3-642-12607-9_5, author={Shivani Sud and Roy Want and Trevor Pering and Kent Lyons and Barbara Rosario and Michelle Gong}, title={Dynamic Migration of Computation through Virtualization of the Mobile Platform}, proceedings={Mobile Computing, Applications, and Services. First International ICST Conference, MobiCASE 2009, San Diego, CA, USA, October 26-29, 2009, Revised Selected Papers}, proceedings_a={MOBICASE}, year={2012}, month={10}, keywords={Virtualization MIDs Intel® AtomTM processor mobile Virtual Machine (VM) live migration}, doi={10.1007/978-3-642-12607-9_5} }- Shivani Sud
Roy Want
Trevor Pering
Kent Lyons
Barbara Rosario
Michelle Gong
Year: 2012
Dynamic Migration of Computation through Virtualization of the Mobile Platform
MOBICASE
Springer
DOI: 10.1007/978-3-642-12607-9_5
Abstract
Virtualization and live migration techniques have long been used in the enterprise server space and have been tuned to address data center usages. These capabilities are now expanding to personal computers including desktops and laptops and more recently into smaller mobile devices such as Netbooks and Mobile Internet Devices (MID). Hardware support for virtualization in these platforms, such as that offered by Intel® AtomTM processor, enables the use of existing operating systems and virtualization software. Our experiments demonstrate that live migration can be used to dynamically offload computation to a nearby desktop computer from a Netbook, taking only 25 seconds over a 100Mbps Ethernet network and approximately 100 seconds over an 802.11n interface with a measured throughput of 70Mbps. Additionally, these experiments highlight the limitations of existing virtualization solutions for migrating computation between small form-factor mobile devices and desktop computers which have widely varying resources and processing capabilities. Finally, we discuss the challenges observed in these early experiments and raise key questions that need to be resolved to enable the design of effective systems that support this use model.