Virtual machine (VM) scheduling is an important technique to efficientlyoperate the computing resources in a data center. Previous work has mainlyfocused on consolidating VMs to improve resource utilization and thus tooptimize energy consumption. However, the interference between collocated VMsis usually ignored, which can result in very worse performance degradation tothe applications running in those VMs due to the contention of the sharedresources. Based on this observation, we aim at designing efficient VMassignment and scheduling strategies where we consider optimizing both theoperational cost of the data center and the performance degradation of runningapplications and then, we propose a general model which captures the inherenttradeoff between the two contradictory objectives. We present offline andonline solutions for this problem by exploiting the spatial and temporalinformation of VMs where VM scheduling is done by jointly consider thecombinations and the life-cycle overlapping of the VMs. Evaluation results showthat the proposed methods can generate efficient schedules for VMs, achievinglow operational cost while significantly reducing the performance degradationof applications in cloud data centers.
展开▼
