公开(公告)号：US11521082B2

公开(公告)日：2022-12-06

申请号：US16688739

申请日：2019-11-19

Applicant: KYNDRYL, INC.

Inventor： Mikael Lindström ,  Daniela Trevisan ,  Alex R. Martin ,  Mu Qiao

IPC: G06N5/04 ,  G06N5/02 ,  G06N20/00 ,  G06F3/06

Abstract: Data protection activity time in a backup environment can be predicted with a computer-implemented method. The method includes gathering datasets of data points from a range of backup components of a backup system and constructing input features for a predictive machine learning model, where the input features relate to backup parameters of the backup system. The method also includes training the predictive machine learning model to predict data protection activity times for the backup system by using the gathered datasets and applying the input features to the model. The method also includes deploying the trained predictive machine learning model to predict a data protection activity time of a backup system formed of a set of the backup components.