Application health monitoring based on historical application health data and application logs
Techniques for monitoring health of an application based on historical application health data and application logs are disclosed. In one embodiment, the historical application health data and the historical application logs associated with a period may be obtained. The application may include multiple services running therein. Priority of services may be determined based on the historical application health data associated with a portion of the period. Priority of exceptions associated with each of the services may be determined based on the historical application health data and the historical application logs associated with the portion of the period. Further, an application regression model may be trained by correlating the priority of the services, the associated priority of the exceptions, and the corresponding historical application health data. The health of the application may be monitored by analyzing real-time application logs using tested application regression model.
1. A method comprising:
obtaining historical application health data and historical application logs associated with an application for a period;
determining priority of services associated with the application based on the historical application health data associated with a portion of the period;
determining priority of exceptions associated with each of the services based on the historical application health data and the historical application logs associated with the portion of the period;
training an application regression model by correlating the priority of the services, the associated priority of the exceptions, and the corresponding historical application health data;
testing the application regression model based on the historical application health data and the historical application logs associated with a remaining portion of the period; and
predicting health of the application for an upcoming period using the application regression model based on the testing.
2. The method of claim 1 , wherein predicting the health of the application comprises:
retrieving real-time application logs associated with the application; and
predicting the health of the application for the upcoming period by analyzing the real-time application logs using the application regression model.
3. The method of claim 1 , wherein testing the application regression model comprises:
predicting health data of the application for the remaining portion of the period using the trained application regression model; and
determining accuracy of the trained application regression model by comparing the predicted health data of the application with the historical application health data associated with the remaining portion of the period, wherein the trained application regression model is used to predict the health data of the application for the upcoming period when the accuracy is greater than or equal to a predefined threshold.
4. The method of claim 3 , further comprising:
retraining the trained application regression model with historical application health data and historical application logs associated with a modified period when the accuracy is less than the predefined threshold.
5. The method of claim 1 , wherein the historical application health data comprise historical services health data of the services, and the historical application logs comprise historical service logs of the services.
6. The method of claim 5 , wherein determining the priority of the services comprises:
analyzing criticality of the services using a trained diagnosis model by correlating the historical application health data with the associated historical services health data; and
determining the priority of the services by assigning weightages to the services based on the criticality of the services.
7. The method of claim 6 , wherein the diagnosis model is trained using supervised machine learning and natural language processing methods.
8. The method of claim 5 , wherein determining the priority of the exceptions associated with each of the services comprises:
determining the exceptions associated with each of the services by analyzing the historical service logs using a trained component model;
analyzing criticality of the exceptions associated with each of the services using the trained component model by correlating each exception with the associated historical services health data and the associated historical application health data; and
determining the priority of the exceptions by assigning weightages to the exceptions based on the criticality of the exceptions.
9. The method of claim 8 , wherein the component model is trained using supervised machine learning and natural language processing methods.
10. A system comprising:
a data training unit is to:
obtain historical application health data and historical application logs associated with an application for a period;
determine priority of services associated with the application based on the historical application health data associated with a portion of the period;
determine priority of exceptions associated with each of the services based on the historical application health data and the historical application logs associated with the portion of the period;
train an application regression model by correlating the priority of the services, the associated priority of the exceptions, and the corresponding historical application health data; and
test the application regression model based on the historical application health data and the historical application logs associated with a remaining portion of the period; and
a real-time application monitoring unit is to predict health of the application for an upcoming period using the application regression model based on the testing.
11. The system of claim 10 , wherein the real-time application monitoring unit is to:
retrieve real-time application logs associated with the application; and
predict the health of the application for the upcoming period by analyzing the real-time application logs using the application regression model.
12. The system of claim 10 , wherein the data training unit is to:
predict health data of the application for the remaining portion of the period using the trained application regression model; and
determine accuracy of the trained application regression model by comparing the predicted health data of the application with the historical application health data associated with the remaining portion of the period, wherein the trained application regression model is used to predict the health data of the application for the upcoming period when the accuracy is greater than or equal to a predefined threshold.
13. The system of claim 12 , wherein the data training unit is to:
retrain the trained application regression model with historical application health data and historical application logs associated with a modified period when the accuracy is less than the predefined threshold.
14. The system of claim 10 , wherein the historical application health data comprise historical services health data of the services, and the historical application logs comprise historical service logs of the services.
15. The system of claim 14 , wherein the data training unit is to:
analyze criticality of the services using a trained diagnosis model by correlating the historical application health data with the associated historical services health data; and
determine the priority of the services by assigning weightages to the services based on the criticality of the services.
16. The system of claim 14 , wherein the data training unit is to:
determine the exceptions associated with each of the services by analyzing the historical service logs using a trained component model;
analyze criticality of the exceptions associated with each of the services using the trained component model by correlating each exception with the associated historical services health data and the associated historical application health data; and
determine the priority of the exceptions by assigning weightages to the exceptions based on the criticality of the exceptions.
17. A non-transitory machine-readable storage medium encoded with instructions that, when executed by a processor, cause the processor to:
obtain historical application health data and historical application logs associated with an application for a period;
determine priority of services associated with the application based on the historical application health data associated with a portion of the period;
determine priority of exceptions associated with each of the services based on the historical application health data and the historical application logs associated with the portion of the period;
train an application regression model by correlating the priority of the services, the associated priority of the exceptions, and the corresponding historical application health data;
test the application regression model based on the historical application health data and the historical application logs associated with a remaining portion of the period; and
predict health of the application for an upcoming period using the application regression model based on the testing.
18. The non-transitory machine-readable storage medium of claim 17 , wherein predicting the health of the application comprises:
retrieving real-time application logs associated with the application; and
predicting the health of the application for the upcoming period by analyzing the real-time application logs using the application regression model.
19. The non-transitory machine-readable storage medium of claim 17 , wherein testing the application regression model comprises:
predicting health data of the application for the remaining portion of the period using the trained application regression model; and
determining accuracy of the trained application regression model by comparing the predicted health data of the application with the historical application health data associated with the remaining portion of the period, wherein the trained application regression model is used to predict the health data of the application for the upcoming period when the accuracy is greater than or equal to a predefined threshold.
20. The non-transitory machine-readable storage medium of claim 19 , further comprising instructions that, when executed by the processor, cause the processor to:
retrain the trained application regression model with historical application health data and historical application logs associated with a modified period when the accuracy is less than the predefined threshold.
21. The non-transitory machine-readable storage medium of claim 17 , wherein the historical application health data comprise historical services health data of the services, and the historical application logs comprise historical service logs of the services.
22. The non-transitory machine-readable storage medium of claim 21 , wherein determining the priority of the services comprises:
analyzing criticality of the services using a trained diagnosis model by correlating the historical application health data with the associated historical services health data; and
determining the priority of the services by assigning weightages to the services based on the criticality of the services.
23. The non-transitory machine-readable storage medium of claim 21 , wherein determining the priority of the exceptions associated with each of the services comprises:
determining the exceptions associated with each of the services by analyzing the historical service logs using a trained component model;
analyzing criticality of the exceptions associated with each of the services using the trained component model by correlating each exception with the associated historical services health data and the associated historical application health data; and
determining the priority of the exceptions by assigning weightages to the exceptions based on the criticality of the exceptions.