A Cloud Resource Allocation Strategy with Entry Control for Multi-priority Cloud Requests.

Considering the complexity and diversity of cloud requests in the cloud environment, cloud requests are divided into primary requests (PRs) with high priority and secondary requests (SRs) with low priority. We introduce the entry threshold and the entry probability to control the entry of SRs. If the number of busy virtual machines (VMs) is greater than or equal to the entry threshold, the recently arrived SR will enter the system with an entry probability that is inversely proportional to the number of busy VMs. Based on the proposed cloud resource allocation strategy with entry control, a discrete-time queueing model is constructed, and the expressions of performance indicators are derived. Compared with the conventional cloud resource allocation strategy without entry control, the proposed cloud resource allocation strategy with entry control can effectively improve the service experience of PRs. For example, when the arrival rate of PRs is equal to 0.5, under the parameter settings given by experiments, the proposed cloud resource allocation strategy can reduce the blocking rate of PRs by 47% and increase the throughput rate of PRs by 55% at most. Finally, we construct a revenue function to obtain the optimal entry thresholds and maximum revenues under different conditions. [ABSTRACT FROM AUTHOR]