Virtual machine capacity planning
Virtual machine capacity planning techniques are disclosed. In various embodiments, a set of time series data is constructed based at least in part on virtual machine related metric values observed with respect to a virtual machine during a training period. The constructed time series data is used to build a forecast model for the virtual machine. The forecast model is used to forecast future values for one or more of the virtual machine related metrics. The forecasted future values are used to determine whether an alert condition is predicted to be met.
1. A method of planning virtual machine capacity, comprising:
constructing a set of time series data based at least in part on values of virtual machine related metrics observed with respect to a virtual machine during a training period to obtain training data;
using the training data to construct a Hotelling T2 chart and compute Hotelling T2 values corresponding to the observed values of each of one or more of the virtual machine related metrics at a predetermined time during a plurality of successive time periods;
using a subset of the virtual machine related metrics of said time series data and the Hotelling T2 chart to build a forecast model for said virtual machine to forecast future values of said one or more of the virtual machine related metrics during a forecast period;
selecting control limits for the Hotelling T2 chart based at least in part on the computed Hotelling T2 values for the training data;
setting an alert threshold based at least in part on the selected control limits and on a difference between said observed values of said subset of virtual machine related metrics and mean values of said subset of virtual machine related metrics for the virtual machine; and
comparing the forecasted values to the alert threshold to predict when an alert condition might be expected to occur in the forecast period,
wherein a responsive action is taken based at least in part in response to determining the alert condition is predicted to be met.
2. The method of claim 1 , wherein the Hotelling T2 values correspond to a distance value associated with a difference between an observed feature vector and an associated mean vector.
3. The method of claim 2 , wherein the distance value is associated with a multivariate normal distribution.
4. The method of claim 1 , wherein the metric values are associated with at least one of health, usage, error, and configuration data.
5. The method of claim 1 , wherein the responsive action includes increasing a capacity of the virtual machine.
6. The method of claim 1 , further comprising receiving the virtual machine related metric values.
7. The method of claim 6 , further comprising determining based at least in part on a statistical analysis of the received virtual machine related metric values a subset of the virtual machine related metric values to be included in a feature set for the virtual machine.
8. The method of claim 7 , further comprising including in the time series data corresponding values for virtual machine metrics included in the feature set.
9. The method of claim 1 , wherein building said forecast model for the virtual machine comprises building a vector auto-regression model for the virtual machine.
10. A virtual machine capacity planning system, comprising:
a processor configured to:
construct a set of time series data based at least in part on values of virtual machine related metrics observed with respect to a virtual machine during a training period to obtain training data;
use the training data to construct a Hotelling T2 chart and compute Hotelling T2 values corresponding to the observed values of each of one or more of the virtual machine related metrics at a predetermined time during a plurality of successive time periods;
use a subset of the virtual machine related metrics of said time series data and the Hotelling T2 chart to build a forecast model for said virtual machine to forecast future values of said one or more of the virtual machine related metrics during a forecast period;
select control limits for the Hotelling T2 chart based at least in part on the computed Hotelling T2 values for the training data;
setting an alert threshold based at least in part on the selected control limits and on a difference between said observed values of said subset of virtual machine related metrics and mean values of said subset of virtual machine related metrics for the virtual machine;
compare the forecasted values to the alert threshold to predict when an alert condition might be expected to occur in the forecast period,
wherein a responsive action is taken based at least in part in response to determining the alert condition is predicted to be met; and
a memory or other storage device coupled to the processor and configured to store the set of time series data.
11. The system of claim 10 , wherein the Hotelling T2 values correspond to a distance value associated with a difference between an observed feature vector and an associated mean vector.
12. The system of claim 10 , wherein the responsive action includes increasing a capacity of the virtual machine.
13. The system of claim 10 , further comprising a communication interface coupled to the processor and configured to receive the virtual machine related metric values.
14. The system of claim 13 , wherein the processor is further configured to determine based at least in part on a statistical analysis of the received virtual machine related metric values a subset of the virtual machine related metric values to be included in a feature set for the virtual machine.
15. The system of claim 14 , wherein the processor is further configured to include in the time series data corresponding values for virtual machine metrics included in the feature set.
16. The system of claim 10 , wherein the processor is configured to use at least in part the constructed time series data to build a vector auto-regression model for the virtual machine.
17. A computer program product to perform virtual machine capacity planning, the computer program product being embodied in a non-transitory computer readable storage medium and comprising computer instructions for:
constructing a set of time series data based at least in part on values of virtual machine related metrics observed with respect to a virtual machine during a training period to obtain training data;
using the training data to construct a Hotelling T2 chart and compute Hotelling T2 values corresponding to the observed values of each of one or more of the virtual machine related metrics at a predetermined time during a plurality of successive time periods;
using a subset of the virtual machine related metrics of said time series data and the Hotelling T2 chart to build a forecast model for said virtual machine to forecast future values of said one or more of the virtual machine related metrics during a forecast period;
selecting control limits for the Hotelling T2 chart based at least in part on the computed Hotelling T2 values for the training data;
setting an alert threshold based at least in part on the selected control limits and on a difference between said observed values of said subset of virtual machine related metrics and mean values of said subset of virtual machine related metrics for the virtual machine; and
comparing the forecasted values to the alert threshold to predict when an alert condition might be expected to occur in the forecast period,
wherein a responsive action is taken based at least in part in response to determining the alert condition is predicted to be met.
18. The computer program product of claim 17 , wherein the Hotelling T2 values correspond to a distance value associated with a difference between an observed feature vector and an associated mean vector.
19. The computer program product of claim 18 , wherein the distance value is associated with a multivariate normal distribution.
20. The computer program product of claim 17 , wherein building said forecast model for the virtual machine comprises building a vector auto-regression model for the virtual machine.