Predicting physical storage unit health
Maintaining a data storage device having a plurality of physical storage units includes constructing a model that predicts failure of the physical storage units in the data storage device, where the model is trained using empirical data containing operational parameters of physical storage units, determining failure predictions of the physical storage units in the data storage device using the model, moving applications in the data storage device based on the failure predictions of the physical storage units in the data storage device, and maintaining the physical storage units in the data storage device according to the failure predictions of the physical storage units in the data storage device. The model may be constructed using a Random Forest Classifier model from the Scikit-Learn software package. Using empirical data containing operational parameters of physical storage units may include accessing publicly available empirical data.
1. A method of maintaining a data storage device having a plurality of physical storage units, comprising:
constructing a model that predicts failure of the physical storage units in the data storage device, wherein the model is trained using empirical data containing operational parameters of physical storage units;
determining failure predictions of the physical storage units in the data storage device using the model;
moving non-critical applications in the data storage device on to ones of the physical storage units that are predicted to fail;
replacing a first set of the physical storage units in the data storage device predicted to fail by a first fixed amount of time; and
scheduling for future replacement a second set of the physical storage units in the data storage device that are predicted to fail after the first fixed amount of time but before a second fixed amount of time.
2. The method of claim 1 , wherein the model is constructed using a Random Forest Classifier model from the Scikit-Learn software package.
3. The method of claim 1 , wherein using empirical data containing operational parameters of physical storage units includes accessing publicly available empirical data.
4. The method of claim 1 , further comprising:
moving applications deemed critical from the physical storage units in the data storage device that are predicted to fail.
5. The method of claim 1 , wherein the physical storage units include disk drives and solid state drives.
6. The method of claim 1 , wherein a host computing device, coupled to the data storage device, determines the failure predictions.
7. The method of claim 6 , wherein the host computing device moves the applications on the data storage device.
8. A method of maintaining a data storage device having a plurality of physical storage units, comprising:
constructing a model that predicts failure of the physical storage units in the data storage device, wherein the model is trained using empirical data containing operational parameters of physical storage units;
determining failure predictions of the physical storage units in the data storage device using the model;
moving applications in the data storage device based on the failure predictions of the physical storage units in the data storage device;
replacing a first set of the physical storage units in the data storage device predicted to fail by a first time; and
scheduling for future replacement a second set of the physical storage units in the data storage device that are predicted to fail after the first time but before a second time, wherein the first time is one month and the second time is two months.
9. A non-transitory computer-readable medium containing software that maintains a data storage device having a plurality of physical storage units, the software comprising:
executable code that constructs a model that predicts failure of the physical storage units in the data storage device, wherein the model is trained using empirical data containing operational parameters of physical storage units;
executable code that determines failure predictions of the physical storage units in the data storage device using the model;
executable code that moves non-critical applications in the data storage device on to ones of the physical storage units that are predicted to fail;
executable code that maintains the physical storage units in the data storage device marks for replacement a first set of the physical storage units in the data storage device predicted to fail by a first fixed amount of time; and
executable code that schedules for future replacement a second set of the physical storage units in the data storage device that are predicted to fail after the first fixed amount of time but before a second fixed amount of time.
10. The non-transitory computer-readable medium of claim 9 , wherein the model is constructed using a Random Forest Classifier model from the Scikit-Learn software package.
11. The non-transitory computer-readable medium of claim 9 , wherein using empirical data containing operational parameters of physical storage units includes accessing publicly available empirical data.
12. The non-transitory computer-readable medium of claim 9 , further comprising:
executable code that moves applications deemed critical from the physical storage units in the data storage device that are predicted to fail.
13. The non-transitory computer-readable medium of claim 9 , wherein the physical storage units include disk drives and solid state drives.
14. The non-transitory computer-readable medium of claim 9 , wherein a host computing device, coupled to the data storage device, includes the executable code that determines the failure predictions.
15. The non-transitory computer-readable medium of claim 14 , wherein the host computing device includes the executable code that moves the applications on the data storage device.
16. A non-transitory computer-readable medium containing software that maintains a data storage device having a plurality of physical storage units, the software comprising:
executable code that constructs a model that predicts failure of the physical storage units in the data storage device, wherein the model is trained using empirical data containing operational parameters of physical storage units;
executable code that determines failure predictions of the physical storage units in the data storage device using the model;
executable code that moves applications in the data storage device based on the failure predictions of the physical storage units in the data storage device;
executable code that maintains the physical storage units in the data storage device marks for replacement a first set of the physical storage units in the data storage device predicted to fail by a first time; and
executable code that schedules for future replacement a second set of the physical storage units in the data storage device that are predicted to fail after the first time but before a second time, wherein the first time is one month and the second time is two months.