Using machine learning to improve input/output performance of an application

Patent metadata

Field	Value
Publication number	US-10771580-B1
Application number	US-201916353153-A
Country	US
Kind code	B1
Filing date	Mar 14, 2019
Priority date	Mar 14, 2019
Publication date	Sep 8, 2020
Grant date	Sep 8, 2020

Abstract

In some examples, a computing device may determine that a selected application is executing and gather, over a predetermined time interval, data associated with operations being performed to the input/output stack by the selected application. After gathering the data, a classifier may analyze the data and determine a particular workload type from a predefined set of workload types associated with the selected application. The computing device may select a particular profile from a plurality of predefined profiles based at least in part on the particular workload type, and modify, based on the particular profile, a plurality of parameters to create a plurality of modified parameters. The modified parameters may reduce an execution time of performing the operations to the input/output stack.

First claim

What is claimed is:

1. A method comprising:
determining, by one or more processors of a computing device, that a selected application is performing operations to an input/output stack of the computing device;
ordering, according to a frequency of occurrence, the operations performed by the selected application to the input/output stack;
determining, by the one or more processors, a subset of the operations comprising a plurality of most frequently performed operations performed by the selected application to the input/output stack;
performing a comparison of the subset of the operations to frequent operations associated with each of a predefined set of workload types;
classifying, by a classifier executed by the one or more processors, the subset of the operations performed by the selected application as a particular workload type from a predefined set of workload types, wherein the classifier comprises a machine learning algorithm trained to classify input/output operations as a type of workload;
selecting, by the one or more processors, a particular profile corresponding to the particular workload type from a plurality of predefined profiles based at least in part on the particular workload type and on the comparison; and
modifying, by the one or more processors and based on the particular profile, a plurality of parameters to create modified parameters to reduce an execution time of the selected application performing the operations to the input/output stack using the modified parameters.

2. The method of claim 1 , wherein modifying the plurality of parameters to create the modified parameters comprises:
modifying a process priority associated with the selected application to a highest priority;
modifying a power plan of an operating system to a high-performance power plan;
modifying a hyperthreading feature associated with the one or more processors;
disabling a core parking feature associated with the one or more processors;
disabling a compression feature to compress data stored in a random-access memory; and
modifying a page combining feature of the operating system to remove duplicates of content stored in the random-access memory.

3. The method of claim 1 , wherein modifying the plurality of parameters to create the modified parameters comprises:
modifying a vertical synchronization feature associated with synchronizing a frame rate output of the selected application with a monitor refresh rate of a display device associated with the computing device; and
modifying a pre-fetch feature associated with the one or more processors to store frequently accessed data in a random-access memory of the computing device.

4. The method of claim 1 , further comprising:
displaying a user interface; and
receiving, via the user interface, a user selection of a particular application of a plurality of applications to create the selected application.

5. The method of claim 1 , wherein before ordering, according to the frequency of occurrence, the operations performed by the selected application to the input/output stack, the method comprises:
gathering over a predetermined interval of time, data associated with the selected application performing the operations to the input/output stack, wherein the predetermined interval of time comprises one of fifteen minutes, thirty minutes, forty-five minutes, or sixty minutes and
performing an analysis of the data.

6. The method of claim 1 , wherein the input/output stack comprises:
a file system used by the computing device;
a random-access memory used by the computing device;
a logical storage used by an operating system;
a cache allocated in the random-access memory by the operating system;
a pagefile used by the operating system; and
physical storage accessible to the operating system.

7. A computing device comprising:
one or more processors; and
one or more non-transitory computer readable media storing instructions executable by the one or more processors to perform operations comprising:
determining that a selected application is performing operations to an input/output stack of the computing device;
ordering, according to a frequency of occurrence, the operations performed by the selected application to the input/output stack;
determining a subset of the operations comprising a plurality of most frequently performed operations performed by the selected application to the input/output stack;
performing a comparison of the subset of the operations to frequent operations associated with each of a predefined set of workload types;
classifying, by a classifier, the subset of the operations as a particular workload type from a predefined set of workload types, wherein the classifier comprises a machine learning algorithm trained to classify input/output operations as a type of workload;
selecting a particular profile corresponding to the particular workload type from a plurality of predefined profiles based at least in part on the particular workload type and on the comparison; and
modifying, based on the particular profile, a plurality of parameters to create modified parameters, the modified parameters reducing an execution time of the selected application performing the operations to the input/output stack using the modified parameters.

8. The computing device of claim 7 , wherein modifying the plurality of parameters to create the modified parameters comprises at least one of:
modifying a process priority associated with the selected application to a highest priority;
modifying a power plan of an operating system to a high-performance power plan;
modifying a hyperthreading feature associated with the one or more processors;
disabling a core parking feature associated with the one or more processors;
disabling a compression feature to compress data stored in a random-access memory; or
modifying a page combining feature of the operating system to remove duplicates of content stored in the random-access memory.

9. The computing device of claim 7 , wherein modifying the plurality of parameters to create the modified parameters comprises:
modifying a vertical synchronization feature associated with synchronizing a frame rate output of the selected application with a monitor refresh rate of a display device associated with the computing device.

10. The computing device of claim 7 , further comprising:
displaying a user interface; and
receiving, via the user interface, a user selection of a particular application of a plurality of applications to create the selected application.

11. The computing device of claim 7 , wherein the input/output stack comprises:
a file system used by the computing device;
a random-access memory used by the computing device;
a logical storage used by an operating system;
a cache allocated in the random-access memory by the operating system;
a pagefile used by the operating system; and
physical storage accessible to the operating system.

12. One or more non-transitory computer readable media storing instructions executable by one or more processors of a computing device to perform operations comprising:
determining that a selected application is performing operations to an input/output stack of the computing device;
ordering, according to a frequency of occurrence, the operations performed by the selected application to the input/output stack;
determining, by the one or more processors, a subset of the operations comprising a plurality of most frequently performed operations performed by the selected application to the input/output stack;
performing a comparison of the subset of the operations to frequent operations associated with each of

Assignees

• Dell Products Lp

Inventors

• Khosrowpour Farzad
• Vichare Nikhil

Classifications

• H04L67/303 Primary

  Terminal profiles · CPC title

• H04L67/564

  Enhancement of application control based on intercepted application data · CPC title

• G06F18/214

  Generating training patterns; Bootstrap methods, e.g. bagging or boosting · CPC title

• G06F18/2185

  the supervisor being an automated module, e.g. intelligent oracle · CPC title

• G06F11/349

  for interfaces, buses · CPC title