This section investigates the downtime minimization
mechanism described in Section 3. From the experiments,
we observed that the average bandwidth values reduced as
the number of host CPU cores on which the VM CPUs
were running reduces. Figure 4 shows TLM live migration
time and downtime performance of the MG, IS, SP, and
BT benchmarks under various parameters. Figure 6 shows downtime values against two CPU
over-committing parameters:the adjust-point and the
reduced-cores (defined in Section 3). Four graphs for the
MG, IS, SP, and BT benchmarks are presented.