A Strategy of Service Quality Optimization for Live Virtual Machine Migration

摘要

Service downtime is inevitable in migrating ongoing virtual machine. It is obvious that service downtime impacts the quality of live virtual machine migration. In this paper, a strategy of reducing downtime by reconfiguring memory space is proposed. It employs algorithm AMP-LvMrt to predict the potential number of inconsistent frames to be duplicated at the stop-and-copy stage followed by performing memory reconfiguration if necessary. Experiments have demonstrated our strategy can effectively shorten downtime to an adequate level without degenerating computation efficiency.