Predictive instance suspension and resumption

Abstract

Remote computing resource service providers allow customers to execute virtual computer systems in a virtual environment on hardware provided by the computing resource service provider. The hardware may be distributed between various geographic locations connected by a network. The distributed environment may increase latency of various operations of the virtual computer systems executed by the customer. To reduce latency of various operations predictive modeling is used to predict the occurrence of various operations and initiate the operations before they may occur, thereby reducing the amount of latency perceived by the customer.

1 - 25 . (canceled) 
     
     
         26 . A computer-implemented method, comprising:
 collecting input data associated with a plurality of instances;   generating, based at least in part on the input data, a predictive model usable to resume an instance, the predictive model including a classifier;   generating a serialization schedule for the instance based at least in part on the predictive model; and   causing a set of operations to be performed to launch the instance based at least in part on the serialization schedule.   
     
     
         27 . The computer-implemented method of  claim 26 , wherein generating the predictive model is based at least in part on a role of the particular virtual machine instance. 
     
     
         28 . The computer-implemented method of  claim 27 , wherein the role corresponds to a service executed by a customer of a computing resource service provider; and
 wherein the input data indicates a first set of intervals of time during which the service is active and a second set of intervals of time during which the service is idle.   
     
     
         29 . The computer-implemented method of  claim 26 , wherein generating the predictive model is based at least in part on a plurality of predictive models. 
     
     
         30 . The computer-implemented method of  claim 26 , wherein the computer-implemented method further comprises seeding a second predictive model based at least in part on the predictive model, the second predictive model associated with a second customer distinct from a customer associated with the predictive model. 
     
     
         31 . The computer-implemented method of  claim 26 , wherein the serialization schedule further comprises an indication of a start time for at least one operation of the set of operations such that the instance is available to a customer prior to a predicted start time, the predicted start time determined based at least in part on the predictive model. 
     
     
         32 . A system, comprising:
 one or more processors; and   memory that stores computer-executable instructions that, if executed, cause the one or more processors to:
 generate a predictive model associated with a first instance, the predictive model generated based at least in part on the input data associated with a plurality of other instances, the predictive model usable to determine a start time of an event for making available the first instance; and 
 cause the first instance to be instantiated by at least initiating one or more operations to make the first instance available prior to the start time. 
   
     
     
         33 . The system of  claim 32 , wherein the input data further comprises information indicating operations of the plurality of other instances initiated at least in part by requests from users. 
     
     
         34 . The system of  claim 32 , wherein the one or more operations includes an operation of loading a portion of a virtual machine image associated with the first instance into memory of a server computer system. 
     
     
         35 . The system of  claim 34 , wherein the start time is determined such that the operation of loading a portion of a virtual machine image is completed prior to the event for making available the first instance. 
     
     
         36 . The system of  claim 32 , wherein the input data further comprises information indicating price information associated with a market of instances. 
     
     
         37 . The system of  claim 32 , wherein the input data further comprises information indicating a first interval of time during which operations were executed by the plurality of other instances and a second interval of time during which the plurality of other instances were idle. 
     
     
         38 . The system of  claim 32 , wherein the memory further includes computer-executable instructions that, if executed, cause the one or more processors to:
 generate a schedule based at least in part on the predictive model, the schedule including the start time; and   cause at least one instance of the plurality of other instances to be instantiated based at least in part on the schedule.   
     
     
         39 . A non-transitory computer-readable storage medium having stored thereon executable instructions that, as a result of being executed by one or more processors of a computer system, cause the computer system to at least:
 obtain input data associated with execution of a plurality of virtual machine instances;   generate a predictive model that indicates a probability of receiving a request to instantiate a particular virtual machine instance of the plurality of virtual machine instances by at least:
 analyzing the input data to generate one or more classifiers of the input data; and 
 generating the predictive model based at least in part on the one or more classifiers; and 
   cause one or more operations involved in instantiating the particular virtual machine instance to occur in accordance with the predictive model.   
     
     
         40 . The non-transitory computer-readable storage medium of  claim 39 , wherein the instructions that cause the computer system to obtain the input data further include instructions that cause the computer system to obtain usage data for the plurality of virtual machine instances. 
     
     
         41 . The non-transitory computer-readable storage medium of  claim 39 , wherein the instructions that cause the computer system to obtain the input data further include instructions that cause the computer system to obtain information indicating a plurality of commands transmitted to a computing resource service provider to perform operations associated with the plurality of virtual machine instances. 
     
     
         42 . The non-transitory computer-readable storage medium of  claim 39 , wherein the instructions further comprise instructions that, as a result of being executed by the one or more processors, cause the computer system to generate one or more additional predictive models based at least in part on obtaining additional input data associated with the plurality of virtual machine instances. 
     
     
         43 . The non-transitory computer-readable storage medium of  claim 42 , wherein the instructions further comprise instructions that, as a result of being executed by the one or more processors, cause the computer system to generate a set of schedules for instantiating virtual machine instances based at least in part on the predictive model and the one or more additional predictive models. 
     
     
         44 . The non-transitory computer-readable storage medium of  claim 43 , wherein the instructions further comprise instructions that, as a result of being executed by the one or more processors, cause the computer system to correlate the set of schedules to determine a start time for causing the one or more operations to occur. 
     
     
         45 . The non-transitory computer-readable storage medium of  claim 39 , wherein the instructions that cause the computer system to obtain the input data further include instructions that cause the computer system to obtain information indicating idle interval of the plurality of virtual machine instances.