Adjusting cloud-based execution environment by neural network

Patent metadata

Field	Value
Publication number	US-10616370-B2
Application number	US-201515753964-A
Country	US
Kind code	B2
Filing date	Aug 21, 2015
Priority date	Aug 21, 2015
Publication date	Apr 7, 2020
Grant date	Apr 7, 2020

Abstract

Various examples provide for solutions that use an artificial neural network to adjust a cloud-based execution environment for an application based on a set of metrics associated with the cloud-based execution environment, including a Quality-of-Service (QoS) metric and a cost metric. The adjustment may be based on a mapping, determined by the artificial neural network, between the application and a cloud-based infrastructure utilized by the cloud-based execution environment.

First claim

The invention claimed is:

1. A method, comprising:
receiving, at a computer system, a set of metrics associated with a cloud-based execution environment for executing an application, the set of metrics including a quality-of-service (QoS) metric, a cost metric, and a performance metric of the application;
determining, by an artificial neural network, a mapping between the application and a cloud-based infrastructure utilized by the cloud-based execution environment to execute the application, by an input node layer of the artificial neural network receiving at least the QoS metric, the cost metric and the performance metric of the application, wherein the mapping identifies a number of virtual machine instances for executing the application, a characteristic of the virtual machine instances, and a manner in which the virtual machine instances are to be utilized by the cloud-based execution environment in order to improve compliance of execution of the application by the cloud-based execution environment with the set of metrics;
adjusting the cloud-based execution environment based on the mapping; and
backpropagating, through the artificial neural network, at least an updated QoS metric and an updated cost metric associated with the adjusted cloud-based execution environment.

2. The method of claim 1 , wherein the mapping comprises an allocation of a set of the number of virtual machine instances to be utilized by the cloud-based execution environment to execute the application.

3. The method of claim 2 , wherein the mapping describes a particular role a particular virtual machine instance in the set of virtual machine instances is to serve in the cloud-based execution environment to execute the application.

4. The method of claim 2 , wherein with respect to a particular virtual machine instance in the set of virtual machine instances, the mapping describes a number of processor cores, amount of memory, a network parameter, or a number logical processors per a network socket.

5. The method of claim 1 , wherein the QoS metric is provided by a service-level agreement (SLA) management tool associated with the cloud-based execution environment.

6. The method of claim 1 , wherein the cost metric is provided by a service-level agreement (SLA) management tool associated with the cloud-based execution environment.

7. The method of claim 1 , wherein the performance metric of the application includes a response time associated with the application, a queue length associated with the application, or computer-resource utilization by the application as the application executes within the cloud-based execution environment.

8. The method of claim 1 , wherein adjusting the cloud-based execution environment based on the mapping comprises provisioning a virtual machine instance for use by the cloud-based execution environment to execute the application, or de-provisioning a virtual machine instance being used by the cloud-based execution environment to execute the application, the provisioning or the de-provisioning being based on the mapping.

9. The method of claim 1 , wherein adjusting the cloud-based execution environment based on the mapping comprises re-configuring an existing virtual machine instance being utilized by the cloud-based execution environment to execute the application.

10. The method of claim 1 , comprising training the artificial neural network by a k-fold cross-validation over a set of records that associate a second set of metrics for a given cloud-based execution environment with a set of parameters for the given cloud-based execution environment.

11. The system of claim 1 , wherein the updated QoS metric represents a difference between the QoS metric and a QoS metric obtained from the adjusted cloud-based execution environment by an SLA management tool associated with the cloud-based execution environment and the updated cost metric represents a difference between the cost metric and a cost metric obtained from the adjusted cloud-based execution environment by the SLA management tool.

12. A computer system, comprising:
a memory that stores a specification describing a quality of service (QoS) parameter and a priority level for individual workloads of server systems;
a processing resource; and
a non-transitory computer-readable medium, coupled to the processing resource, having stored therein instructions that when executed by the processing resource cause the processing resource to:
obtain from a service level agreement (SLA) management tool a set of metrics associated with a cloud-based execution environment for executing an application, the SLA management tool being associated with the cloud-based execution environment, and the set of metrics including a quality-of-service (QoS) metric and a cost metric;
obtain performance metrics of the application;
determine, based on an artificial neural network, a mapping between the application and a cloud-based infrastructure utilized by the cloud-based execution environment to execute the application, an input node layer of the artificial neural network receiving at least the QoS metric, the cost metric, and the performance metric, wherein the mapping identifies a number of virtual machine instances for executing the application, a characteristic of the virtual machine instances, and a manner in which the virtual machine instances are to be utilized by the cloud-based execution environment in order to improve compliance of execution of the application by the cloud-based execution environment with the set of metrics;
train the artificial neural network by a k-fold cross-validation over a set of records that associate a second set of metrics for a given cloud-based execution environment with a set of parameters for the given cloud-based execution environment;
adjust the cloud-based execution environment based on the mapping; and
backpropagate, through the artificial neural network, at least an updated QoS metric and an updated cost metric associated with the adjusted cloud-based execution environment.

13. The system of claim 12 , wherein adjusting the cloud-based execution environment based on the mapping comprises re-configuring an existing virtual machine instance being utilized by the cloud-based execution environment to execute the application.

14. The system of claim 12 , wherein adjusting the cloud-based execution environment based on the mapping comprises provisioning, based on the mapping, a new virtual machine instance for use by the cloud-based execution environment to execute the application, or de-provisioning, based on the mapping, an existing virtual machine instance being used by the cloud-based execution environment to execute the application.

15. A non-transitory computer readable medium having instructions stored thereon, the instructions being executable by a processor of a computer system, the instructions causing the processor to:
obtain, from a SLA management tool, a set of metrics associated with a cloud-based execution environment a set of metrics associated with a cloud-based execution environment for executing an application, the SLA management tool being associated with the cloud-based execution environment, and the set of metrics including a quality-of-service (QoS) metric and a cost metric;
obtain a performance metric of the application;
determine, by an artificial neural network, a mapping between the application and a cloud-based infrastructure utilized by the cloud-based execution environment to execute the application, an input node layer of the artificial neural network receiving at least the QoS metric, the cost metric, and the perform

Assignees

• Hewlett Packard Entpr Dev Lp

Inventors

• Bardhan Soumendu

Classifications

• G06F9/45558

  Hypervisor-specific management and integration aspects · CPC title

• G06N3/084

  Backpropagation, e.g. using gradient descent · CPC title

• H04L41/5003

  Managing SLA; Interaction between SLA and QoS · CPC title

• G06F2009/45595

  Network integration; Enabling network access in virtual machine instances · CPC title

• G06N3/02

  Neural networks · CPC title