Cloud data center load aware routing request method based on SDN

Abstract

The invention relates to a cloud data center load aware routing request method based on SDN. The cloud data center load aware routing request method based on SDN comprehensively takes into account the network congestion information and the virtual machine load information to obtain the best virtual machine to process an application request and the path combination to arrive at the virtual machine, so as to minimize the response time. In the process, the response time calculation model of a load aware controller is proposed; in addition, the link set through which each application can send the request to the virtual machine is obtained based on the genetic algorithm, the calculation model for calculating the required time for each virtual machine to process the current application request is provided, and the calculation model of the delay time of the request in the link is given, so as to obtain the minimized total response time finally by calculating the time that the virtual machine processes the request and the delay time of the link. By means of the invention, the delay time of the application in the data center is reduced, the user experience is improved, and the loss of the profit for the cloud data center provider is reduced.

Description

technical field [0001] The invention relates to an SDN-based cloud data center load-aware request routing method. Background technique [0002] Physical servers in cloud data center infrastructure are virtualized and pooled into multiple isolated virtual machines. The requests accepted by the cloud data center first need to go through the cloud data center network, and then reach the underlying virtual machine for processing. Therefore, the total latency of a request is the sum of the latency of the request in the network and in the virtual machine. Studies have shown that the delay time of applications in cloud data centers has a greater effect on user experience, and will bring profit losses to cloud data center providers. [0003] The recently emerging software-defined network technology can provide centralized and unified control of the network through programmable network devices based on the OpenFlow protocol. These network devices include software-defined network (...

AI Technical Summary

Problems solved by technology

However, the total latency of a request is the sum of the latency of the request on the network and in the virtual machine, so the processing latency in the underlying virtual machine also has a significant impact on the user experience

If the request is routed to an abnormal or overloaded virtual machine, then this will cause a large virtual machine processing time

In addition, the routing mechanism in the existing controller is usually relatively simple, such as Floodlight only selects the target virtual machine according to the principle of the shortest path

This simple routing method may cause large network delays in the network due to network congestion or overloaded virtual machines or large processing delays in virtual machines, which may result in large overall response times

Images