US 12,008,407 B2
Automatically identifying and right sizing instances
Brian Toal, San Francisco, CA (US); and Manpreet Singh, Hyderabad (IN)
Assigned to Salesforce, Inc., San Francisco, CA (US)
Filed by Salesforce, Inc., San Francisco, CA (US)
Filed on Jun. 30, 2022, as Appl. No. 17/854,652.
Application 17/854,652 is a continuation of application No. 16/566,209, filed on Sep. 10, 2019, granted, now 11,379,266.
Prior Publication US 2022/0350663 A1, Nov. 3, 2022
This patent is subject to a terminal disclaimer.
Int. Cl. G06F 9/50 (2006.01)
CPC G06F 9/5027 (2013.01) [G06F 9/5083 (2013.01); G06F 2209/501 (2013.01); G06F 2209/5011 (2013.01); G06F 2209/505 (2013.01)] 24 Claims
OG exemplary drawing
 
1. A system to provide a compute optimization service for recommending compute resource usage, the system comprising:
memory circuitry to store program code of a resource analyzer, an instance type determiner, and an instance type recommender; and
processor circuitry connected to the memory circuitry, wherein:
the processor circuitry is to operate the resource analyzer to analyze resource utilization metrics of a set of resources belonging to a set of instances, the resource utilization metrics corresponding to a first level of performance or cost;
the processor circuitry is to operate the instance type determiner to determine, based on the resource utilization metrics, a recommended instance type of the set of instances that is predicted to provide at least a second cost that is more optimal than the first cost; and
the processor circuitry is to operate the instance type recommender to:
cause evaluation of the at least one instance having the recommended instance type when the recommended instance type is different from a current instance type of the at least one instance, and
provide, based on the evaluation, a recommendation to facilitate a replacement or resizing of the at least one instance having the current instance type with the at least one instance having the recommended instance type,
wherein the processor circuitry is configured to cause evaluation by an ensemble of machine learning (ML) models based on the resource utilization metrics to determine workload patterns of the at least one instance,
wherein the ensemble of ML models further determines the recommendation based on a current workload data of the at least one instance, the determined workload patterns, and a range of desired performances and costs,
wherein, based on the recommendation, the at least one instance having the current instance type is replaced with or resized based on the at least one instance having the recommended instance type, and
wherein one or more workloads of the at least one instance are executed using the recommended instance type.