Predictive scaling of computing resources

Abstract

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.

What is claimed is: 
     
         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, at a first time, to serve responses to client requests for data via a first set of provisioned computing resources; 
 predicting a need for a second set of provisioned computing resources to be operational at a second time that is later than the first time, in which the second set of computing resources is modified relative to the first set of computing resources; 
 requesting provisioning of a modified amount of computing resources to result in the second set of computing resources, the requesting of the provisioning of the modified amount computing resources performed at a prediction time that is earlier than the second time; and 
 operating, at the second time, to serve responses to client requests for data via the second set of provisioned computing resources. 
   
     
     
         2 . The system of  claim 1 , wherein the operations further comprise building an offline schedule that comprises respective needed provisioned resource counts and respective prediction times, including the prediction time that is earlier than the second time. 
     
     
         3 . The system of  claim 2 , wherein the operations further comprise modifying the schedule based upon a constraint. 
     
     
         4 . The system of  claim 2 , wherein the operations further comprise optimizing the schedule. 
     
     
         5 . The system of  claim 2 , wherein the offline schedule is based on at least one of: historical data or override data. 
     
     
         6 . The system of  claim 1 , wherein the modified amount of the computing resources increases a first number of computing resources of the first set to result in a second number of computing resources of the second set, to allow the second set of computing resources to be operational by the second time, and wherein the operations further comprise determining the prediction time based on a computing resources spin-up time. 
     
     
         7 . The system of  claim 6 , wherein the computing resources spin-up time is a function of a difference between the second number of computing resources and the first number of computing resources, and wherein the operations further comprise obtaining or determining the computing resources spin-up time based on the difference. 
     
     
         8 . The system of  claim 1 , wherein the operations further comprise dynamically building a schedule that comprises one or more needed provisioned resource counts and a prediction time for each needed provisioned resource count, wherein the requesting of the provisioning of the modified amount of computing resources is based on the schedule. 
     
     
         9 . The system of  claim 1 , wherein the modified amount of the computing resources decreases a first number of computing resources of the first set to result in a second number of computing resources of the second set. 
     
     
         10 . The system of  claim 1 , wherein the first set of computing resources comprises a first subgroup of computing resources corresponding to a first service, wherein the first set of computing resources comprises a second subgroup of computing resources corresponding to a second service, and wherein the requesting of the provisioning of the modified amount of computing resources to result in the second set of computing resources comprises requesting a first modified amount of computing resources to modify the first subgroup, and requesting a second modified amount of computing resources to modify the second subgroup. 
     
     
         11 . The system of  claim 1 , wherein the computing resources are provisioned by a cloud computing system. 
     
     
         12 . A method, comprising:
 obtaining, by a system comprising a processor, an amount of computing resources needed to serve responses to client data requests at a future time, in which the amount of computing resources is increased relative to a currently operating set of resources; and   requesting, by the system, provisioning of an increased amount of computing resources, to proactively provision the increased amount of computing resources at a prediction time that is prior to the future time and in which the prediction time is based at least in part on an amount of time needed for the increased amount of computing resources to become operational.   
     
     
         13 . The method of  claim 12 , further comprising generating, by the system, a schedule comprising first data representing the prediction time and second data representing the increased amount of computing resources, wherein the obtaining of the amount of computing resources comprises accessing the schedule. 
     
     
         14 . The method of  claim 13 , wherein the generating of the schedule is based on at least one of: historical data, override data, constraint data, or optimization data. 
     
     
         15 . The method of  claim 12 , further comprising receiving, by the system, an event associated with the increased amount, and wherein the requesting of the provisioning occurs in response to the event. 
     
     
         16 . A non-transitory machine-readable medium, comprising executable instructions that, when executed by a processor, facilitate performance of operations, the operations comprising:
 determining a provisioning time window corresponding to an amount of time that a group of computing resources takes to become operational;   obtaining a prediction time based upon the provisioning time window and a predicted need for the group of computing resources to be operational by an operating time; and   requesting provisioning of the group of computing resources at a request time that is based on the provisioning time window and the operating time, resulting in the group of computing resources being operational by the operating time.   
     
     
         17 . The non-transitory machine-readable medium of  claim 16 , wherein the operations further comprise determining the operating time based on historical data. 
     
     
         18 . The non-transitory machine-readable medium of  claim 16 , wherein the group of computing resources comprises a first subgroup corresponding to a first service and a second subgroup corresponding to a second service, and wherein the requesting of the provisioning comprises requesting a first amount of computing resources for the first subgroup and a second amount of computing resources for the second subgroup. 
     
     
         19 . The non-transitory machine-readable medium of  claim 16 , wherein the request time is a first request time, wherein the group of computing resources comprises a first group, and wherein the operations further comprise requesting deprovisioning of a second group of computing resources at a second request time, resulting in an amount of computing resources of the second group being decreased relative to the amount of computing resources of the first group. 
     
     
         20 . The non-transitory machine-readable medium of  claim 16 , wherein the obtaining of the prediction time comprises at least one of: accessing a schedule, or receiving an event.