Confident peak-aware response time estimation by exploiting telemetry data from different system configurations
A prediction manager for providing responsiveness predictions for deployments includes persistent storage and a predictor. The persistent storage stores training data and conditioned training data. The predictor is programmed to obtain training data based on: a configuration of at least one deployment of the deployments, and a measured responsiveness of the at least one deployment, perform a peak extraction analysis on the measured responsiveness to obtain conditioned training data, obtain a prediction model using: the training data, and a first untrained prediction model, obtain a confidence prediction model using: the conditioned training data, and a second untrained prediction model, obtain a combined prediction using: the prediction model, and the confidence prediction model, and perform, based on the combined prediction, an action set to prevent a responsiveness failure.
1. A prediction manager for providing responsiveness predictions for deployments, comprising:
storage for storing training data;
a predictor programmed to:
obtain the training data based on:
a configuration of at least one deployment of the deployments, and
a measured responsiveness of the at least one deployment;
perform a peak extraction analysis on the measured responsiveness to obtain conditioned training data;
obtain a prediction model using:
the training data, and
a first untrained prediction model;
obtain a confidence prediction model using:
the conditioned training data, and
a second untrained prediction model;
obtain a combined prediction using:
the prediction model, and
the confidence prediction model, wherein the combined prediction specifies:
a predicted response time for storage resources of the at least one deployment, and
at least one confidence indicator for the predicted response time;
make a determination that the at least one confidence indicator falls below a minimum confidence threshold; and
perform, based on the combined prediction determination, an action set to prevent a responsiveness failure.
2. The prediction manager of claim 1 , wherein the prediction model relates hypothetical configurations of the deployments and the responsiveness predictions.
3. The prediction manager of claim 2 , wherein the confidence prediction model relates the hypothetical configurations and confidence indicators associated with the responsiveness predictions.
4. The prediction manager of claim 3 , wherein the combined prediction is obtained using a hypothetical configuration of the hypothetical configurations as input to:
the prediction model, and
the confidence prediction model.
5. The prediction manager of claim 1 , wherein performing the peak extraction analysis on the measured responsiveness to obtain the conditioned training data comprises:
identifying at least one feature in the measured responsiveness that is obscured from the first untrained prediction model; and
adding the at least one feature to the conditioned training data.
6. A method for providing responsiveness predictions for deployments, comprising:
obtaining training data based on:
a configuration of at least one deployment of the deployments, and
a measured responsiveness of the at least one deployment;
performing a peak extraction analysis on the measured responsiveness to obtain conditioned training data;
obtaining a prediction model using:
the training data, and
a first untrained prediction model;
obtaining a confidence prediction model using:
the conditioned training data, and
a second untrained prediction model;
obtaining a combined prediction using:
the prediction model, and
the confidence prediction model, wherein the combined prediction specifies:
a predicted response time for storage resources of the at least one deployment, and
at least one confidence indicator for the predicted response time;
making a determination that the at least one confidence indicator falls below a minimum confidence threshold; and
performing, based on the combined prediction determination, an action set to prevent a responsiveness failure.
7. The method of claim 6 , wherein the prediction model relates hypothetical configurations of the deployments and responsiveness predictions.
8. The method of claim 7 , wherein the confidence prediction model relates the hypothetical configurations and confidence indicators associated with the responsiveness predictions.
9. The method of claim 8 , wherein the combined prediction is obtained using a hypothetical configuration of the hypothetical configurations as input to:
the prediction model, and
the confidence prediction model.
10. The method of claim 6 , wherein performing the peak extraction analysis on the measured responsiveness to obtain the conditioned training data comprises:
identifying at least one feature in the measured responsiveness that is obscured from the first untrained prediction model; and
adding the at least one feature to the conditioned training data.
11. A non-transitory computer readable medium comprising computer readable program code, which when executed by a computer processor enables the computer processor to perform a method for providing responsiveness predictions for deployments, the method comprising:
obtaining training data based on:
a configuration of at least one deployment of the deployments, and
a measured responsiveness of the at least one deployment;
performing a peak extraction analysis on the measured responsiveness to obtain conditioned training data;
obtaining a prediction model using:
the training data, and
a first untrained prediction model;
obtaining a confidence prediction model using:
the conditioned training data, and
a second untrained prediction model;
obtaining a combined prediction using:
the prediction model, and
the confidence prediction model, wherein the combined prediction specifies:
a predicted response time for storage resources of the at least one deployment, and
at least one confidence indicator for the predicted response time;
making a determination that the at least one confidence indicator falls below a minimum confidence threshold; and
performing, based on the combined prediction determination, an action set to prevent a responsiveness failure.
12. The non-transitory computer readable medium of claim 11 , wherein the prediction model relates hypothetical configurations of the deployments and responsiveness predictions.
13. The non-transitory computer readable medium of claim 12 , wherein the confidence prediction model relates the hypothetical configurations and confidence indicators associated with the responsiveness predictions.
14. The non-transitory computer readable medium of claim 13 , wherein the combined prediction is obtained using a hypothetical configuration of the hypothetical configurations as input to:
the prediction model, and
the confidence prediction model.