US 11,809,910 B2
System and method for dynamically resizing computational infrastructure to accommodate unexpected demands
Naga Vamsi Krishna Akkapeddi, Charlotte, NC (US)
Assigned to Bank of America Corporation, Charlotte, NC (US)
Filed by BANK OF AMERICA CORPORATION, Charlotte, NC (US)
Filed on Oct. 14, 2020, as Appl. No. 17/070,195.
Prior Publication US 2022/0114020 A1, Apr. 14, 2022
Int. Cl. G06F 9/50 (2006.01); G06N 20/00 (2019.01); G06F 11/34 (2006.01); G06F 16/21 (2019.01)
CPC G06F 9/505 (2013.01) [G06F 9/5027 (2013.01); G06F 11/3433 (2013.01); G06F 16/219 (2019.01); G06N 20/00 (2019.01); G06F 9/50 (2013.01); G06F 9/5011 (2013.01); G06F 9/5016 (2013.01); G06F 9/5044 (2013.01)] 17 Claims
OG exemplary drawing
 
7. A method comprising:
detecting that an event has occurred;
in response to detecting that the event has occurred:
applying a machine learning algorithm to predict, based on the occurrence of the event, that a future value of a resource usage of a first computational resource at a future time will be greater than a maximum value of the resource usage of the first computational resource, wherein:
the first computational resource has a capacity and is associated with the resource usage, the capacity of the first computational resource corresponding to the maximum value of the resource usage of the first computational resource;
the machine learning algorithm is trained, based on training data, to predict, based on the occurrence of the event, that the future value of the resource usage of the first computational resource at the future time will be greater than the maximum value of the resource usage of the first computational resource, wherein the training data comprises:
a set of historical data for a first subsystem, wherein:
the first subsystem comprises the first computational resource; and
the set of historical data comprises values of the resource usage of the first computational resource over a period of time; and
a first set of historical events, each historical event of the first set of historical events occurring at an event time and associated with an increase in the resource usage of the first computational resource occurring at a time later than the event time;
prior to the future time, increasing the capacity of the first computational resource such that the increased capacity of the first computational resource corresponds to a new maximum value of the resource usage of the first computational resource, the new maximum value greater that the future value;
wherein:
a first application executing on the first subsystem and a second application executing on the first subsystem share the first computational resource;
the values of the resource usage of the first computational resource over the period of time comprise:
first values of the resource usage of the first computational resource by the first application over the period of time; and
second values of the resource usage of the first computational resource by the second application over the period of time;
the machine learning algorithm is further trained, based on the set of historical data and the first set of historical events, to predict, based on the occurrence of a second event, that a second future value of the resource usage of the first computational resource by the first application at a second future time will be greater than a threshold; and
the method further comprises:
detecting that the second event has occurred; and
in response to detecting that the second event has occurred:
applying the machine learning algorithm to predict, based on the occurrence of the second event, that the second future value of the resource usage of the first computational resource by the first application at the second future time will be greater than the threshold; and
transmitting an alert to the second application, the alert indicating that the resource usage of the first computational resource by the first application is expected to increase.