Capacity analysis using closed-system modules
A capacity-analysis tool (CAT) provides a model framework for creating a model of a capacity-planning-target (CPT) system, e.g., a data center. The tool includes a model framework that, in turn, includes a closed-system template for creating CSMs, i.e., models of capacity-limited systems. A user uses the CAT to create CPT models using the CSMs as building blocks. A machine-learning engine is used to train the CPT model, converting parameter time-series data to functions of time. The trained CPT models are then used to make capacity-planning estimates, e.g., time remaining on a system before usage matches capacity. The CAT makes it easy to extend a model, e.g., by adding new dimensions (new factors of interest) in the form of new CSMs to which the new dimensions have been assigned.
1. A process comprising:
creating, via a processor, a capacity-planning target (CPT) model of a CPT system, the CPT model comprising a plurality of closed-system models (CSMs), wherein each CSM is assigned a respective dimension and a plurality of parameters for the respective dimension, the plurality of parameters comprising a capacity parameter, the plurality of parameters comprising at least one of a usage parameter or a demand parameter, at least one of the parameters corresponding to a respective time series of datapoints, wherein the CPT model specifies a hierarchical relationship at least between a first respective CSM and a second respective CSM of the plurality of CSMs;
training, via the processor, the CPT model to convert the respective time-series of datapoints to a respective function of time, the respective function of time thereby becoming part of the trained CPT model;
making, using the trained CPT model, capacity-analysis estimates based at least in part on the respective function of time, wherein the trained CPT model makes the capacity-analysis estimates based on the hierarchical relationship between the first respective CSM and the second respective CSM; and
determining, via the processor, a plurality of estimated time remaining values for the plurality of CSMs for an instance in which the demand parameter or the usage parameter meets the capacity parameter based at least in part on the capacity-analysis estimates.
2. The process of claim 1 , wherein making the capacity-analysis estimates involves evaluating the respective function of time for a future time.
3. The process of claim 1 , wherein making the capacity-analysis estimates involves combining plural functions of time across components and within a dimension to yield a system function of time for the dimension.
4. The process of claim 3 , wherein making the capacity-analysis estimates further involves making estimates based at least in part on the system function of time.
5. The process of claim 4 , wherein [[the]] making the capacity-analysis estimates further involves combining the estimates based at least in part on the system functions of time to yield the capacity-analysis estimates.
6. The process of claim 1 , further comprising:
training the CPT model with a particular CSM to yield a retrained CPT model; and
making new capacity-analysis estimates using the retrained CPT model with the particular CSM.
7. The process of claim 1 , further comprising:
generating, using the trained CPT model, a correlation model of the CPT model, wherein the correlation model comprises a plurality of correlations among the plurality of parameters.
8. The process of claim 1 , further comprising:
determining, using the CPT model, a shortest time remaining before the usage parameter or the demand parameter meets the capacity parameter based on the plurality of estimated time remaining values for the plurality of CSMs.
9. A capacity-analysis system comprising:
a computing device that includes a processor;
a program, when executed by the processor, causes the computing device to at least:
create a capacity planning target (CPT) model of a CPT system, the CPT model having a plurality of closed-system models (CSMs) associated therewith a dimension role to which a dimension can be assigned and a plurality of parameter roles to which a plurality of parameters can be assigned, the plurality of parameters including a capacity parameter and a demand parameter;
train the CPT model by converting time-series data associated with at least one of the plurality of parameters to at least one respective function of time based on a hierarchical relationship among the plurality of CSMs; and
determine, via the processor, a plurality of estimated time remaining values for the plurality of CSMs for an instance in which the demand parameter meets the capacity parameter based at least in part on a plurality of capacity-analysis estimates made by the CPT model.
10. The capacity-analysis system of claim 9 , wherein the program is further configured to generate functions of time for some of the plurality of parameters based on functions of time generated from the time-series data.
11. The capacity-analysis system of claim 10 , wherein the CPT model includes a version of an algorithm for making the plurality of capacity-analysis estimates.
12. The capacity-analysis system of claim 11 , wherein the program further causes the computing device to create the algorithm based on the plurality of parameters being assigned to the plurality of parameter roles.
13. The capacity analysis system of claim 12 , wherein the algorithm provides for computing the plurality of capacity-analysis estimates for what-if scenarios provided to the CPT model.
14. The capacity-analysis system of claim 9 , wherein each CSM determines a future time in which the demand parameter exceeds the capacity parameter for the dimension.
15. A system, comprising:
a computing device that includes a processor;
a program that, when executed by the processor, causes the computing device to at least:
create a capacity planning target (CPT) model of a CPT system, the CPT model including a plurality of closed-system models (CSMs), wherein each CSM is assigned a respective dimension and a plurality of parameters for the respective dimension, the plurality of parameters comprising a capacity parameter and at least one of a usage parameter or a demand parameter, at least one of the plurality of parameters corresponding to a respective time series of datapoints, wherein the CPT model specifies a hierarchical relationship at least between a first respective CSM and a second respective CSM of the plurality of CSMs;
train the CPT model to convert the respective time-series of datapoints to a respective function of time, the respective function of time thereby becoming part of the CPT model;
make capacity-analysis estimates, using the trained CPT model, based on the respective function of time, wherein the trained CPT model makes the capacity-analysis estimates based on the hierarchical relationship between the first respective CSM and the second respective CSM; and
determine a plurality of estimated time remaining values for the plurality of CSMs for an instance in which the demand parameter or the usage parameter meets the capacity parameter based least in part on the capacity-analysis estimates.
16. The system of claim 15 wherein making the capacity-analysis estimates includes evaluating the respective function of time for a future time.
17. The system of claim 15 wherein making the capacity-analysis estimates includes combining plural functions of time across components and within a dimension to yield a system function of time for the dimension.
18. The system of claim 17 , wherein making the capacity-analysis estimates further involves making estimates based at least in part on the system function of time.
19. The system of claim 18 , wherein making the capacity-analysis estimates further includes combining the estimates based on system functions of time to yield the capacity-analysis estimates.
20. The system of claim 15 , wherein the CPT model is in a form of an eXtensible Markup Language (XML) document.