Granted Patent B2

US 12,430,587 · App. 17/074,179 · Granted Sep 30, 2025

Inventors: Shanand Reddy (Kulim, MY); Chew Chien Wern (Penang, MY); Choo Lead Ta (Puchong, MY)

Assignee: Dell Products L.P.

G06N20/10G06F16/21

Abstract

Claims (56)

1. A computer-implemented method comprising:

determining one or more first predicted growth values of a first object in a database comprising a plurality of objects, using a first non-linear regression machine learning model, wherein the first non-linear regression machine learning model is trained, using a first set of data records indicative of a size of the first object over time, to generate a confidence threshold band around the one or more first predicted growth values;

automatically updating at least a portion of the first set of data records by labeling a given data record as anomalous in response to an actual growth value of the first object deviating from the confidence threshold band generated by the first non-linear regression machine learning model corresponding to one or more second predicted growth values of the first object output by the first non-linear regression machine learning model;

training a first classification model based at least in part on the updated data records in the first set of data records, wherein the first classification model is trained to identify a threshold indicative of abnormal growth of the first object that is indicative of one or more database performance issues;

monitoring the size of the first object using the first non-linear regression machine learning model and the first classification model based on one or more additional data records indicative of the size of the first object;

predicting, based on the monitoring, one or more of the database performance issues, in response to detecting that at least one third predicted growth value of the first object, output by the first non-linear regression machine learning model for at least one of the additional data records, exceeds the threshold;

performing one or more automated maintenance operations on the database to preemptively address at least a portion of the one or more predicted database performance issues, wherein the one or more automated maintenance operations comprise an index creation operation that programmatically generates a data index and integrates the data index into the database to reduce data search and retrieval times within the database as compared to data search and retrieval times without the generated data index; and

separately monitoring at least one second object of the plurality of objects in the database using a second non-linear regression machine learning model and a second classification model that are trained based at least in part on a second set of data records indicative of a size of the at least one second object over time;

wherein the method is performed by at least one processing device comprising a processor coupled to a memory.

2. The computer-implemented method of claim 1 , wherein the first non-linear regression machine learning model comprises a Gaussian process regression model.

3. The computer-implemented method of claim 1 , wherein the first classification model comprises a support vector machine model.

4. The computer-implemented method of claim 1 , wherein each of the data records in the first set comprises information corresponding to at least one of:

a delete operation;

an insert operation;

a current size of the first object;

a difference in size of the first object over a period of time;

a time; and

a date.

5. The computer-implemented method of claim 1 , wherein the one or more automated maintenance operations further comprise at least one of an archive operation and a purge operation on the database.

6. The computer-implemented method of claim 1 , wherein the first set of data records comprises historical database records corresponding to the first object for a given period of time.

7. The computer-implemented method of claim 1 , wherein the training comprises adjusting at least one training parameter based on a designated threshold of accuracy for a predicted growth of the first object.

8. The computer-implemented method of claim 1 , wherein the first object comprises a data table, and wherein the data index categorizes data records in the data table based on one or more fields to enable retrieval of a given data record in the database without searching other portions of the data table.

9. The computer-implemented method of claim 1 , wherein the one or more automated maintenance operations further comprise a query tuning operation that adjusts at least one of a syntax and a clause arrangement of a query corresponding to the first object.

10. A non-transitory processor-readable storage medium having stored therein program code of one or more software programs, wherein the program code when executed by at least one processing device causes the at least one processing device:

to determine one or more first predicted growth values of a first object in a database comprising a plurality of objects, using a first non-linear regression machine learning model, wherein the first non-linear regression machine learning model is trained, using a first set of data records indicative of a size of the first object over time, to generate a confidence threshold band around the one or more first predicted growth values;

to automatically update at least a portion of the first set of data records by labeling a given data record as anomalous in response to an actual growth value of the first object deviating from the confidence threshold band generated by the first non-linear regression machine learning model corresponding to one or more second predicted growth values of the first object output by the first non-linear regression machine learning model;

to train a first classification model based at least in part on the updated data records in the first set of data records, wherein the first classification model is trained to identify a threshold indicative of abnormal growth of the first object that is indicative of one or more database performance issues;

to monitor the size of the first object using the first non-linear regression machine learning model and the first classification model based on one or more additional data records indicative of the size of the first object;

to predict, based on the monitoring, one or more of the database performance issues, in response to detecting that at least one third predicted growth value of the first object, output by the first non-linear regression machine learning model for at least one of the additional data records, exceeds the threshold;

to perform one or more automated maintenance operations on the database to preemptively address at least a portion of the one or more predicted database performance issues, wherein the one or more automated maintenance operations comprise an index creation operation that programmatically generates a data index and integrates the data index into the database to reduce data search and retrieval times within the database as compared to data search and retrieval times without the generated data index; and

to separately monitor at least one second object of the plurality of objects in the database using a second non-linear regression machine learning model and a second classification model that are trained based at least in part on a second set of data records indicative of a size of the at least one second object over time.

11. The non-transitory processor-readable storage medium of claim 10 , wherein the first non-linear regression machine learning model comprises a Gaussian process regression model.

12. The non-transitory processor-readable storage medium of claim 10 , wherein the first classification model comprises a support vector machine model.

13. The non-transitory processor-readable storage medium of claim 10 , wherein each of the data records in the first set comprises information corresponding to at least one of:

a delete operation;

an insert operation;

a current size of the first object;

a difference in size of the first object over a period of time;

a time; and

a date.

14. The non-transitory processor-readable storage medium of claim 10 , wherein the one or more automated maintenance operations further comprise at least one of an archive operation and a purge operation on the database.

15. The non-transitory processor-readable storage medium of claim 10 , wherein the one or more automated maintenance operations further comprise a query tuning operation that adjusts at least one of a syntax and a clause arrangement of a query corresponding to the first object.

16. An apparatus comprising:

at least one processing device comprising a processor coupled to a memory;

the at least one processing device being configured:

to determine one or more first predicted growth values of a first object in a database comprising a plurality of objects, using a first non-linear regression machine learning model, wherein the first non-linear regression machine learning model is trained, using a first set of data records indicative of a size of the first object over time, to generate a confidence threshold band around the one or more first predicted growth values;

to automatically update at least a portion of the first set of data records by labeling a given data record as anomalous in response to an actual growth value of the first object deviating from the confidence threshold band generated by the first non-linear regression machine learning model corresponding to one or more second predicted growth values of the first object output by the first non-linear regression machine learning model;

to train a first classification model based at least in part on the updated data records in the first set of data records, wherein the first classification model is trained to identify a threshold indicative of abnormal growth of the first object that is indicative of one or more database performance issues;

to monitor the size of the first object using the first non-linear regression machine learning model and the first classification model based on one or more additional data records indicative of the size of the first object;

to predict, based on the monitoring, one or more of the database performance issues, in response to detecting that at least one third predicted growth value of the first object, output by the first non-linear regression machine learning model for at least one of the additional data records, exceeds the threshold;

to perform one or more automated maintenance operations on the database to preemptively address at least a portion of the one or more predicted database performance issues, wherein the one or more automated maintenance operations comprise an index creation operation that programmatically generates a data index and integrates the data index into the database to reduce data search and retrieval times within the database as compared to data search and retrieval times without the generated data index; and

to separately monitor at least one second object of the plurality of objects in the database using a second non-linear regression machine learning model and a second classification model that are trained based at least in part on a second set of data records indicative of a size of the at least one second object over time.

17. The apparatus of claim 16 , wherein the first non-linear regression machine learning model comprises a Gaussian process regression model.

18. The apparatus of claim 16 , wherein the first classification model comprises a support vector machine model.

19. The apparatus of claim 16 , wherein the one or more automated maintenance operations further comprise at least one of an archive operation and a purge operation on the database.

20. The apparatus of claim 16 , wherein the one or more automated maintenance operations further comprise a query tuning operation that adjusts at least one of a syntax and a clause arrangement of a query corresponding to the first object.

Assignments (9)

Continuity (1)

References Cited (6)