Technique selection for file system utilization prediction

Patent metadata

Field	Value
Publication number	US-12380061-B2
Application number	US-202016863640-A
Country	US
Kind code	B2
Filing date	Apr 30, 2020
Priority date	Apr 30, 2020
Publication date	Aug 5, 2025
Grant date	Aug 5, 2025

Abstract

A trained classification model is executed, causing a classification of a first set of file system usage data into a set of categories comprising a trend category and a periodicity category. Responsive to the first set of file system usage data being classified into the trend category, a time series of the first set of file system usage data is generated. Responsive to the first set of file system usage data being classified into the periodicity category, using an anomaly detection model, an anomaly within the first set of file system usage data is detected. Responsive to predicting that the time series will exceed a threshold, a first reconfiguring of a file system resource is caused, altering a capacity of the file system. Responsive to detecting the anomaly, a second reconfiguring of the file system resource is caused, altering a capacity of the file system.

First claim

What is claimed is:

1. A computer-implemented method comprising:
executing, using a processor and a memory, a trained classification model, the executing causing a classification of a first set of file system usage data into a set of categories, the set of categories comprising a trend category, a flat category, and a periodicity category, the first set of file system usage data comprising usage data of a file system on at least one data storage media device, wherein the flat category classification describes a condition in response to the first set of file system usage data remaining above a first threshold and below a second threshold of a set of predefined thresholds;
causing, using the processor and the memory, at least one operation from a set of operations, the set of operations comprising (i) generating, responsive to the first set of file system usage data being classified into the trend category, a time series of the first set of file system usage data, (ii) detecting, responsive to the first set of file system usage data being classified into the periodicity category, using an anomaly detection model, an anomaly within the first set of file system usage data, and (iii) determining, responsive to the first set of file system data usage being classified into the flat category, that that no changes to the file system are necessary; and
causing, using the processor and the memory, when the category is not the flat category, at least one alteration from a set of alterations, the set of alterations comprising (i) responsive to predicting that the time series will exceed the first threshold of the set of predefined thresholds, a first reconfiguring of a file system resource, the first reconfiguring altering a capacity of the file system by adding processing and network bandwidth to meet a processing and network bandwidth forecast associated with the time series, and (ii) responsive to detecting the anomaly, a second reconfiguring of the file system resource, the second reconfiguring altering a capacity of the file system by adding processing and network bandwidth to meet a processing and network bandwidth forecast associated with the anomaly.

2. The computer-implemented method of claim 1 , further comprising:
adjusting, using a second set of file system usage data, the trained classification model, the second set of file system usage data comprising file system usage data collected subsequent to the first reconfiguring, or the second reconfiguring, or a no change state for the flat category.

3. The computer-implemented method of claim 1 , further comprising: adjusting, using a second set of file system usage data, the first threshold of the set of predefined thresholds, the second set of file system usage data comprising file system usage data collected subsequent to the first reconfiguring.

4. The computer-implemented method of claim 1 , further comprising: adjusting, using a second set of file system usage data, the first reconfiguring, the second set of file system usage data comprising file system usage data collected subsequent to the first reconfiguring.

5. The computer-implemented method of claim 1 , further comprising: adjusting, using a second set of file system usage data, the anomaly detection model, the second set of file system usage data comprising file system usage data collected subsequent to the second reconfiguring.

6. The computer-implemented method of claim 1 , further comprising: adjusting, using a second set of file system usage data, the second reconfiguring, the second set of file system usage data comprising file system usage data collected subsequent to the second reconfiguring.

7. The computer-implemented method of claim 1 , further comprising: causing, using the processor and the memory, a third alteration, the third alteration comprising, responsive to predicting that the time series will go below the second threshold of the set of predefined thresholds, a third reconfiguration of the file system, the third reconfiguring altering a capacity of the file system on the storage device by reducing processing and network bandwidth.

8. The computer-implemented method of claim 1 , further comprising:
responsive to predicting that the time series will go below the first threshold, reallocating a portion of file system capacity to another use, wherein processing and network bandwidth are reallocated to the other use,
wherein the trained classification model comprises a neural network classification model trained using a training set, the training set comprising file system usage data classified into a category in the set of categories.

9. A computer program product for file system utilization prediction, the computer program product comprising:
one or more computer readable storage media, and program instructions collectively stored on the one or more computer readable storage media, the program instructions comprising:
program instructions to execute, using a processor and a memory, a trained classification model, the executing causing a classification of a first set of file system usage data into a set of categories, the set of categories comprising a trend category, a periodicity category, and a flat category, the first set of file system usage data comprising usage data of a file system on a data storage device, wherein the flat category classification describes a condition in response to the first set of file system usage data remaining above a first threshold and below a second threshold of a set of predefined thresholds, the file system comprising at least one storage media device managing files and filenames in data structures, the at least one storage media device comprising at least one of a local data storage device and a remote storage device accessible over a network;
program instructions to cause, using the processor and the memory, at least one operation from a set of operations, the set of operations comprising (i) generating, responsive to the first set of file system usage data being classified into the trend category, a time series of the first set of file system usage data, (ii) detecting, responsive to the first set of file system usage data being classified into the periodicity category, using an anomaly detection model, an anomaly within the first set of file system usage data, and (iii) determining, responsive to the first set of file system data usage being classified into the flat category, that no changes to the file system are necessary; and
program instructions to cause, using the processor and the memory, when the category is not the flat category, at least one alteration from a set of alterations, the set of alterations comprising (i) responsive to predicting that the time series will exceed the first threshold of the set of predefined thresholds, a first reconfiguring of a file system resource, the first reconfiguring altering a capacity of the file system by adding one or more virtual machines to meet a virtual machines forecast associated with the time series, and (ii) responsive to detecting the anomaly, a second reconfiguring of the file system resource, the second reconfiguring altering a capacity of the file system by adding one or more virtual machines to meet a virtual machines forecast associated with the anomaly.

10. The computer program product of claim 9 , further comprising: program instructions to adjust, using a second set of file system usage data, the trained classification model, the second set of file system usage data comprising file system usage data collected subsequent to the first reconfiguring, or the second reconfiguring, or a no change state for the flat category.

11. The computer program product of claim 9 , further comprisin

Assignees

• Kyndryl Inc

Inventors

• Shanmugam Sundaravelu
• Venkatesan Vidhya Shankar
• Mahamuni Amol Bhaskar

Classifications

• G06N3/09

  Supervised learning · CPC title

• G06N3/0442

  characterised by memory or gating, e.g. long short-term memory [LSTM] or gated recurrent units [GRU] · CPC title

• G06N3/08

  Learning methods · CPC title

• G06F16/1734

  Details of monitoring file system events, e.g. by the use of hooks, filter drivers, logs · CPC title

• G06N3/045

  Combinations of networks · CPC title