Apparatus and method to predict a time interval taken for a live migration of a virtual machine

An apparatus comprising:

• a processor coupled to a memory and configured to:
• extract, for each of virtual machines, a first time at which a first live migration has been performed and a first time-interval that has been taken for the first live migration, from log information which stores events of the first live migration,
• extract, for each of the virtual machines, load information from a load history in which the load information including a CPU usage rate and a memory usage amount is stored at predetermined intervals for each of the virtual machines,
• generate a model that predicts a second time-interval to be taken for a second live migration that is expected to be performed for each of the virtual machines, from the load information at the first time and the first time-interval,
• upon reception of an instruction for predicting the second time-interval to be taken for the second live migration, predict the second time-interval from the model, and
• provide the predicted second time-interval to be taken for the second live migration;
• and
• the memory configured to store the log information and the load history,
• when maintenance is performed on a first physical machine in a predetermined period, the second time-interval to be taken for the second live migration of each of first virtual machines mounted on the first physical machine is predicted at each of time points in the predetermined period by using the model corresponding to the each of the first virtual machines;
• and
• an order of the second live migrations of the first virtual machines in the predetermined period is scheduled so that a total of the second time-intervals to be taken for the second live migrations of the first virtual machines is smallest,
• when the scheduled order of the second live migrations of the first virtual machines is predicted to affect an operation of a user of a second virtual machine, information to call attention to the user is provided on a screen for the user of the second virtual machine.

A method comprising:

• extracting, for each of virtual machines, a first time at which a first live migration has been performed and a first time-interval that has been taken for the first live migration, from log information which stores events of the first live migration;
• extracting, for each of the virtual machines, load information from a load history in which the load information including a CPU usage rate and a memory usage amount is stored at predetermined intervals for each of the virtual machines;
• generating a model that predicts a second time-interval to be taken for a second live migration that is expected to be performed for each of the virtual machines, from the load information at the first time and the first time-interval;
• upon receiving an instruction for predicting the second time-interval to be taken for the second live migration, predicting the second time-interval from the model;
• and
• providing the predicted second time-interval to be taken for the second live migration,
• when maintenance is performed on a first physical machine in a predetermined period, the second time-interval to be taken for the second live migration of each of first virtual machines mounted on the first physical machine is predicted at each of time points in the predetermined period by using the model corresponding to the each of the first virtual machines;
• and
• an order of the second live migrations of the first virtual machines in the predetermined period is scheduled so that a total of the second time-intervals to be taken for the second live migrations of the first virtual machines is smallest,
• when the scheduled order of the second live migrations of the first virtual machines is predicted to affect an operation of a user of a second virtual machine, information to call attention to the user is provided on a screen for the user of the second virtual machine.