Prediction and repair of database fragmentation

Patent metadata

Field	Value
Publication number	US-11210274-B2
Application number	US-201916396103-A
Country	US
Kind code	B2
Filing date	Apr 26, 2019
Priority date	Apr 26, 2019
Publication date	Dec 28, 2021
Grant date	Dec 28, 2021

Abstract

Methods, information handling systems and computer readable media are disclosed for detection and repair of fragmentation in databases. In one embodiment, a method includes obtaining log data reflecting transactions in a database, where the log data is generated during operation of the database. The method continues with applying a machine learning classification model to at least a portion of the log data to obtain a first prediction, where the first prediction indicates whether defragmentation of the database should be scheduled. In this embodiment the method also includes using a machine learning time series forecasting model to obtain a second prediction, where the second prediction identifies a future time interval of low relative database utilization, and scheduling a defragmentation procedure for performance during the future time interval of low relative database utilization.

First claim

What is claimed is:

1. A method, comprising:
obtaining log data reflecting transactions in a database, wherein the log data is generated during operation of the database;
applying a machine learning classification model to at least a portion of the log data to obtain a first prediction, wherein the first prediction indicates whether defragmentation of the database should be scheduled;
using a machine learning time series forecasting model to obtain a second prediction, wherein the second prediction identifies a future time interval of low relative database utilization; and
scheduling a database defragmentation procedure for performance during the future time interval of low relative database utilization.

2. The method of claim 1 , wherein using the machine learning time series forecasting model is performed subsequent to obtaining the first prediction indicating that defragmentation of the database should be performed.

3. The method of claim 1 , wherein the first prediction does not indicate that defragmentation should be scheduled, further comprising repeating the obtaining the log data and applying the machine learning classification model.

4. The method of claim 1 , wherein the first prediction indicates that defragmentation should be scheduled, further comprising executing one or more queries of the database for verification of the presence of fragmentation.

5. The method of claim 4 , wherein verification of the presence of fragmentation is not obtained via executing the one or more queries, further comprising adding the at least a portion of the log data and an indication of a non-fragmented database state to a collection of development data for the machine learning classification model.

6. The method of claim 1 , further comprising, subsequent to obtaining the second prediction, executing one or more queries of the database for verification or correction of the future time of low relative database utilization, and wherein scheduling the database defragmentation procedure is performed subsequent to the verification or correction of the future time of low relative database utilization.

7. The method of claim 1 , wherein using the machine learning time series forecasting model comprises:
obtaining a set of utilization metric values for the database, wherein the set of utilization metric values comprises values of a utilization metric for a series of time intervals approaching the time of the obtaining the set of utilization metric values; and
applying the machine learning time series forecasting model to at least a portion of the set of utilization metric values.

8. An information handling system comprising:
one or more processors;
a non-transitory computer-readable storage medium coupled to the one or more processors; and
a plurality of instructions, encoded in the computer-readable storage medium and configured to cause the one or more processors to
obtain log data reflecting transactions in a database, wherein the log data is generated during operation of the database,
apply a machine learning classification model to at least a portion of the log data to obtain a first prediction, wherein the first prediction indicates whether defragmentation of the database should be scheduled,
use a machine learning time series forecasting model to obtain a second prediction, wherein the second prediction identifies a future time interval of low relative database utilization, and
schedule a database defragmentation procedure for performance during the future time interval of low relative database utilization.

9. The information handling system of claim 8 , wherein the plurality of instructions is further configured to use the machine learning time series forecasting model subsequent to obtaining a first prediction indicating that defragmentation of the database should be performed.

10. The information handling system of claim 8 , wherein the plurality of instructions is further configured to, if the first prediction does not indicate that defragmentation should be scheduled, repeat obtaining log data and applying the machine learning classification model.

11. The information handling system of claim 8 , wherein the plurality of instructions is further configured to, if the first prediction indicates that defragmentation should be scheduled:
execute one or more queries of the database for verification of the presence of fragmentation, and
use the machine learning time series forecasting model subsequent to obtaining the verification of the presence of fragmentation.

12. The information handling system of claim 11 , wherein the plurality of instructions is further configured to, if verification of the presence of fragmentation is not obtained via executing the one or more queries, add the at least a portion of the log data and an indication of a non-fragmented database state to a collection of development data for the machine learning classification model.

13. The information handling system of claim 8 , wherein the plurality of instructions is further configured to, subsequent to obtaining the second prediction:
execute one or more queries of the database for verification or correction of the future time of low relative database utilization, and
schedule the database defragmentation procedure subsequent to the verification or correction of the future time of low relative database utilization.

14. The information handling system of claim 8 , wherein the plurality of instructions is further configured to, as part of using the machine learning time series model:
obtain a set of utilization metric values for the database, wherein the set of utilization metric values comprises values of a utilization metric for a series of time intervals approaching the time of obtaining the set of utilization metric values; and
apply the machine learning time series forecasting model to at least a portion of the set of utilization metric values.

15. A non-transitory computer readable storage medium having program instructions encoded therein, wherein the program instructions are executable to:
obtain log data reflecting transactions in a database, wherein the log data is generated during operation of the database;
apply a machine learning classification model to at least a portion of the log data to obtain a first prediction, wherein the first prediction indicates whether defragmentation of the database should be scheduled;
use a machine learning time series forecasting model to obtain a second prediction, wherein the second prediction identifies a future time interval of low relative database utilization; and
schedule a database defragmentation procedure for performance during the future time interval of low relative database utilization.

16. The non-transitory computer readable storage medium of claim 15 , wherein the program instructions are further executable to use the machine learning time series forecasting model subsequent to obtaining a first prediction indicating that defragmentation of the database should be performed.

17. The non-transitory computer readable storage medium of claim 15 , wherein the program instructions are further executable to, if the first prediction indicates that defragmentation should be scheduled:
execute one or more queries of the database for verification of the presence of fragmentation, and
use the machine learning time series forecasting model subsequent to obtaining the verification of the presence of fragmentation.

18. The non-transitory computer readable storage medium of claim 17 , wherein the program instructions a

Assignees

• Dell Products Lp

Inventors

• Sukumaran Shanand Reddy
• Hor Ching Siew
• Palo Ashish Kumar

Classifications

• G06F16/1734

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

• G06F16/217 Primary

  Database tuning (G06F16/2282 takes precedence; database performance monitoring G06F11/3409) · CPC title

• G06F16/1724 Primary

  Details of de-fragmentation performed by the file system (saving storage space on storage systems G06F3/0608; management of blocks in storage devices G06F3/064) · CPC title