Secure configuration in a virtualised computing environment

摘要

A machine learning algorithm is trained as a classifier based on a plurality of training data items, each training data item corresponding to a training virtual machine (VM) and including a representation of parameters for a configuration of the training VM and a representation of characteristics of security attacks for the training VM. One or more relationships are identified between VM configuration parameters and attack characteristics by sampling the trained machine learning algorithm. A set of VM configuration parameters are identified that are indicated as being associated with a particular attack characteristic. A VM configuration for a target VM is then generated, wherein the configuration parameters in the identified set are absent in the generated VM configuration. The generated VM configuration may be applied to the target VM, or the existing target VM configuration may be adapted in accordance with the generated VM configuration. The machine learning algorithm may be a restricted Boltzmann machine, and may include a plurality of hidden units and a plurality of visible units.