Scheduling server maintenance using machine learning
A method and apparatus using machine learning for scheduling server maintenance. In one embodiment of the method, load values for a server are recorded over a period of time, wherein each of the load values is time stamped with a date and time. A first plurality of the load values are classified. The classified first plurality of values are then processed to create a model for predicting a load value of the server. The model is used to generate a first predicted load value of the server for a first date and a first time.
1. A method comprising:
collecting load values for a server over a period of time, wherein each of the load values is time stamped with a date and time at which a respective load value of the load values was collected;
classifying each of a first plurality of the load values;
process the classified first plurality of values to create a model for predicting a load value of the server;
using the model to generate a first predicted load value of the server for a first date and a first time.
2. The method of claim 1 further comprising:
dividing the load values into the first plurality of load values and a second plurality of load values;
comparing the first predicted load value with a load value of the second plurality that is time stamped with the first date and the first time.
3. The method of claim 1 further comprising:
dividing the load values into the first plurality of values and a second plurality of load values;
using the model to generate predicted load values for respective time stamps of the second plurality of load values, wherein the first predicted load value is one of the predicted load values, and wherein the first date and the first time is the time stamp of one of the second plurality of load values;
comparing the predicted load values with respective load values of the second plurality.
4. The method of claim 1 wherein the classified first plurality of values are processed using a support vector machine to create the model.
5. The method of claim 1 further comprising:
using the model to generate predicted load values for respective times of a day, wherein the first predicted load value is one of the predicted load values, and wherein the first time is one of the times of the day;
comparing the predicted load values to each other to identify an optimal load value from the predicted load values;
scheduling the server for maintenance at the time of the day that corresponds to the optimal load value.
6. The method of claim 1 wherein the load values comprise central processing unit (CPU) load values for the server with each CPU load value expressed as a percentage.
7. The method of claim 1 wherein the load values comprise central processing unit (CPU) load values for the server with each CPU load value expressed as a percentage, memory load values for the server with each memory load value expressed as a percentage, disk load values for the server with each disk load value expressed as a percentage, and network load values for the server with each network load value expressed as a percentage.
8. The method of claim 1 wherein the load values comprise central processing unit (CPU) load values for the server with each CPU load value expressed as a percentage, memory load values for the server with each memory load value expressed as a percentage, disk load values for the server with each disk load value expressed as a percentage, or network load values for the server with each network load value expressed as a percentage.
9. The method of claim 1 wherein each of the first plurality of the load values is classified as low or high.
10. A non-transitory computer readable media storing instructions executable by one or more processors to perform a method, the method comprising:
accessing load values measured at a server over a period of time, wherein each of the load values is time stamped with a date and time at which a respective load value of the load values was collected;
classifying each of a first plurality of the load values;
process the classified first plurality of values to create a model for predicting a load value of the server;
using the model to generate a first predicted load value of the server for a first date and a first time.
11. The non-transitory computer readable media of claim 10 , wherein the method further comprises:
dividing the load values into the first plurality of values and a second plurality of load values;
using the model to generate predicted load values for respective time stamps of the second plurality of load values, wherein the first predicted load value is one of the predicted load values, and wherein the first date and the first time is the time stamp of one of the second plurality of load values;
comparing the predicted load values with respective load values of the second plurality.
12. The non-transitory computer readable of claim 10 wherein the classified first plurality of values are processed using a support vector machine to create the model.
13. The non-transitory computer readable method of claim 10 , wherein the method further comprises:
using the model to generate predicted load values for respective times of a day, wherein the first predicted load value is one of the predicted load values, and wherein the first time is one of the times of the day;
comparing the predicted load values to each other to identify an optimal load value from the predicted load values;
scheduling the server for maintenance at the time of the day that corresponds to the optimal load value.
14. The non-transitory computer readable of claim 10 wherein the load values comprise central processing unit (CPU) load values for the server with each CPU load value expressed as a percentage, memory load values for the server with each memory load value expressed as a percentage, disk load values for the server with each disk load value expressed as a percentage, and network load values for the server with each network load value expressed as a percentage.
15. The non-transitory computer readable of claim 10 wherein each of the first plurality of the load values is classified as low or high.
16. An apparatus comprising:
one or more processors; and
one or more non-transitory computer-readable media storing instructions executable by the one or more processors to a method, the method comprising:
accessing load values measured at a server over a period of time, wherein each of the load values is time stamped with a date and time at which a respective load value of the load values was collected;
classifying each of a first plurality of the load values;
process the classified first plurality of values to create a model for predicting a load value of the server;
using the model to generate a first predicted load value of the server for a first date and a first time.
17. The apparatus of claim 16 , wherein the method further comprises:
dividing the load values into the first plurality of values and a second plurality of load values;
using the model to generate predicted load values for respective time stamps of the second plurality of load values, wherein the first predicted load value is one of the predicted load values, and wherein the first date and the first time is the time stamp of one of the second plurality of load values;
comparing the predicted load values with respective load values of the second plurality.
18. The apparatus of claim 16 wherein the classified first plurality of values are processed using a support vector machine to create the model.
19. The apparatus of claim 16 , wherein the method further comprises:
using the model to generate predicted load values for respective times of a day, wherein the first predicted load value is one of the predicted load values, and wherein the first time is one of the times of the day;
comparing the predicted load values to each other to identify an optimal load value from the predicted load values;
scheduling the server for maintenance at the time of the day that corresponds to the optimal load value.
20. The apparatus of claim 16 wherein the load values comprise central processing unit (CPU) load values for the server with each CPU load value expressed as a percentage, memory load values for the server with each memory load value expressed as a percentage, disk load values for the server with each disk load value expressed as a percentage, and network load values for the server with each network load value expressed as a percentage.