Predictive scaling of computing resources
The described technology is generally directed towards predicting a need for provisioned (e.g., cloud computing) resources so that the provisioned resources are proactively provisioned and operational before their actual need. Historical data is used to predict upcoming resource demand so that resources begin to be scaled up to meet demand in advance, at a prediction time, given that there is a “spin-up” delay between requesting resources and their actual availability. The predictive resource provisioning is adaptable to override customary historical data with expected exceptions, and is flexible to handle variable spin-up times, constraints, and optimizations.
1. A system, comprising:
a processor; and
a memory that stores executable instructions that, when executed by the processor, facilitate performance of operations, the operations comprising:
operating a service that handles data traffic via a first number of provisioned computing resources of a cloud computing system;
determining a resource provisioning time, corresponding to a predicted change in data traffic load to be handled by the service;
determining a prediction time, based on the resource provisioning time and a resource spin-up time; and
sending, before the prediction time, simulated traffic that causes the cloud computing system to change the first number of provisioned computing resources to a second number of provisioned computing resources to handle the predicted change in the data traffic load by the resource provisioning time.
2. The system of claim 1 , wherein the operations further comprise determining a value corresponding to the second number of provisioned computing resources based on the predicted change in the data traffic load.
3. The system of claim 1 , wherein the operations further comprise determining the resource spin-up time based on the predicted change in the data traffic load.
4. The system of claim 1 , wherein the resource spin-up time is based on the second number of provisioned computing resources.
5. The system of claim 1 , wherein the determining the resource provisioning time and the determining the prediction time comprise generating a schedule, and wherein the requesting, at the prediction time, the cloud computing system to change the first number of provisioned computing resources to the second number of provisioned computing resources comprises reading the schedule.
6. The system of claim 5 , wherein the generating the schedule comprises using override data corresponding to an expected exception.
7. The system of claim 5 , wherein the generating the schedule comprises applying a constraint.
8. The system of claim 5 , wherein the operations further comprise applying an optimization to the schedule.
9. The system of claim 1 , wherein the operations further comprise, determining a resource deprovisioning time, and, based on the resource deprovisioning time, requesting the cloud computing system to change the second number of provisioned computing resources to a third number of provisioned computing resources.
10. The system of claim 1 , wherein the resource spin-up time is based at least in part on at least one physical computing machine.
11. One or more non-transitory machine-readable storage media having machine-executable instructions, which when executed cause a processor to perform operations, the operations comprising:
facilitating provisioning a first number of computing resources;
determining a predicted change in data traffic load at an upcoming time based at least in part on historical data;
determining, based on the predicted change in data traffic load at an upcoming time, a computing resource provisioning time;
determining, based on the predicted change in the data traffic load at the upcoming time, a second number of computing resources that is different from the first number of computing resources to handle the predicted change in the data traffic load;
determining a prediction time, based on the computing resource provisioning time and a resource spin-up time;
sending, before the prediction time, simulated traffic that causes changing of the first number of provisioned computing resources to a second number of provisioned computing resources to handle the predicted change in the data traffic load by the resource provisioning time; and
facilitating provisioning the second number of computing resources starting based on the prediction time.
12. The one or more non-transitory machine-readable storage media of claim 11 , wherein the determining the computing resource provisioning time, determining the second number of computing resources, and the determining the prediction time comprises generating a schedule.
13. The one or more non-transitory machine-readable storage media of claim 11 , wherein the determining the second number of computing resources, and the determining the computing resource provisioning time, comprises analyzing historical data and state data.
14. A method, comprising:
operating, by a system comprising a processor, a service that handles data traffic via a first number of provisioned computing resources of a cloud computing system;
determining, by the system, a resource provisioning time corresponding to a predicted change in data traffic load to be handled by the service is determined;
determining, by the system, a prediction time based on the resource provisioning time and a resource spin-up time is determined;
sending, by the system, before the prediction time, simulated traffic to the cloud computing system that causes the first number of provisioned computing resources to change to a second number of provisioned computing resources to handle the predicted change in the data traffic load by the resource provisioning time.
15. The method of claim 14 , wherein the determining the resource provisioning time and the determining the prediction time comprise generating a schedule.
16. The method of claim 15 , wherein the generating the schedule comprises using override data corresponding to an expected exception.
17. The method of claim 16 , wherein the override data is manually configured for a predicted exception to the predicted change in data traffic based upon historical data.
18. The method of claim 15 , wherein the generating the schedule comprises applying a constraint.
19. The method of claim 18 , wherein the constraint is a limit on a number of possible changes in the data traffic load to be handled by the service within a certain timeframe.
20. The method of claim 14 , wherein the determining the second number of provisioned computing resources, and the determining the resource provisioning time, comprises analyzing prior traffic patterns, current time of day and day of week, and geographical information.