Safety container placement method and system based on deep reinforcement learning

Abstract

The invention relates to a security container placement method and system based on deep reinforcement learning, and the method comprises the steps: S1, receiving a distribution request sequence of a to-be-distributed container, and initializing a placement strategy of the to-be-distributed container; s2, inputting the state of the current data center into a strategy neural network, outputting a current placement strategy, executing action, and updating the state of the data center after placing one container in the request sequence at one working node; s3, the step S2 is repeated, the next container in the distribution request sequence is sequentially placed, and after all the containers in the distribution request sequence are placed, the container co-residence rate, the user load balancing index value and the reward are calculated; taking the rewards as feedback to guide the next round of training and updating network parameters; and S4, repeating the steps S1 to S3 according to a preset round, and obtaining an action sequence with the highest reward in each round as a placement strategy. According to the method provided by the invention, the probability of container co-residence among different users can be reduced while the load balance of the users is ensured, and the security threat among the containers is relieved.

Description

technical field [0001] The present invention relates to the fields of cloud computing and machine learning, in particular to a method and system for placing secure containers based on deep reinforcement learning. Background technique [0002] In the field of cloud computing, lightweight virtualization technology represented by container technology has risen strongly in recent years. Container is a lightweight operating system layer virtualization technology in the kernel. In the Linux kernel, it is mainly implemented by two mechanisms: Namespace and Cgroup. Compared with traditional virtualization technology, which uses a closed Guest OS to isolate the processes of different users, containers, a lightweight virtualization technology that only relies on the underlying operating system kernel software isolation mechanism, start faster and have higher resource utilization. However, while container technology triggers a new information technology revolution, it also introduces ...

AI Technical Summary

Problems solved by technology

[0006] CN106897110B proposes a container scheduling method and a management node scheduler, but this method only considers performance issues such as resource allocation and system consumption, and does not enhance the security of containers through placement strategies

CN105938437B provides a virtual machine deployment method that is resistant to co-location in a cloud environment, but this invention only considers the deployment of virtual machines on physical machines, and does not consider the scenario where containers are deployed on virtual machine nodes in cloud services

Images