Research on two-layer dynamically deployable virtual machine scheduling strategy

Research on two-layer dynamically deployable virtual machine scheduling strategy

Jiang, East China University of Technology

This paper proposes a new virtual machine scheduling framework to allocate the resources of cloud computing data center to users reasonably and efficiently. This scheduling strategy has made innovations in the following three areas. 1. Introduction of two-tier queuing mechanism. The first level queue is the license queue, which controls the user's access rights, and the second ranks are classified as processing queues, and the final processing priority is determined according to the type of user, the queue time and the proportion of the type of user. The Fs-aloha algorithm is used to detect and handle collisions to ensure that all users ' requests are responded to in a timely manner. If the user revokes the request, it can be logged out directly in the queue to avoid repetitive operation and improve the efficiency of the system execution. 2. The introduction of the monitor to the underlying virtual machine and physical server monitoring. Because the user type and the job type are different, so the monitor's monitoring index is not same, this article proposed seven kinds of monitors. This real-time monitoring can dynamically detect the running state of each index in the system. At the same time, the Ant colony optimization Search algorithm is deployed in the monitoring, and the implementation ability and processing state of each task processor are mastered accurately and in real time by this heuristic algorithm. The user submits a request only through the monitor's instructions, can find the suitable processor, but does not need to search individually. 3. The introduction of the routing keyword effectively binds the queue and the monitor. Because of the different types of users, the jobs that need to be handled are different, so different user jobs need to monitor different metrics. After the user submits the request, the system labels each request (that is, the routing keyword). This new virtual machine scheduling framework combines the queue, routing keyword and monitor to form an efficient and secure scheduling framework (QRM framework). Through theoretical analysis, the two-layer dynamically deployable virtual machine scheduling model can reduce the system error rate while increasing the scheduling efficiency.

Research on two-layer dynamically deployable virtual machine scheduling strategy