Technologies for allocating resources across data centers
Technologies for allocating resources across data centers include a compute device to obtain resource utilization data indicative of a utilization of resources for a managed node to execute a workload. The compute device is also to determine whether a set of resources presently available to the managed node in a data center in which the compute device is located satisfies the resource utilization data. Additionally, the compute device is to allocate, in response to a determination that the set of resources presently available to the managed node does not satisfy the resource utilization data, a supplemental set of resources to the managed node. The supplemental set of resources are located in an off-premises data center that is different from the data center in which the compute device is located. Other embodiments are also described and claimed.
1 . A computing system configurable to be used with local data center resources in association with providing of at least one cloud-based service, the at least one cloud-based service being associated with execution of at least one workload, the local data center resources comprising local physical rack server resources that are to communicate with the computing system via communication links, the computing system comprising:
other data center resources remote from the local data center resources, the other data center resources being configurable to comprise compute resources, storage resources, and accelerator resources; and
management resources comprising circuitry, the management resources being configurable for use in allocating and/or deallocating the compute resources, the storage resources, the accelerator resources, and/or the local physical rack server resources for use in the execution of the at least one workload;
wherein:
the at least one workload comprises at least one virtual machine workload and/or at least one container workload;
the compute resources, the storage resources, and the accelerator resources are associated with at least one cloud service provider;
the local physical rack server resources are at one or more premises associated with a customer of the at least one cloud service provider;
the management resources are configurable to dynamically allocate and/or dynamically deallocate, on an as-needed basis, based upon (1) telemetry-based present resource utilization data, (2) future resource utilization prediction data, (3) application programming interface (API) data, (4) resource utilization trend data, and/or (5) quality of service-related data, the compute resources, the storage resources, the accelerator resources, and/or the local physical rack server resources for use in the execution of the at least one workload;
the accelerator resources and the local physical rack server resources are configurable to comprise graphics processing unit circuits for use in the execution, at least in part, of the at least one workload;
the at least one workload is configurable to perform at least one machine learning-related operation;
the management resources are to select, based upon the API data, the compute resources, the storage resources, the accelerator resources, and/or the local physical rack server resources for dynamic allocation and/or dynamic deallocation for use in the execution of the at least one workload; and
the API data is configurable to be communicated to the management resources in accordance with multiple different protocols.
2 . The computing system of claim 1 , wherein:
the local physical rack server resources are configurable to comprise physical rack server compute resources, physical rack server storage resources, and physical rack server accelerator resources for use in the execution, at least in part, of the at least one workload.
3 . The computing system of claim 2 , wherein:
the management resources are to communicate, at least in part, via an API.
4 . The computing system of claim 3 , wherein:
the management resources are configurable to allocate and/or deallocate the compute resources, the storage resources, the accelerator resources, and/or the local physical rack server resources, based upon client request data and resource utilization cost data.
5 . The computing system of claim 4 , wherein:
the local data center resources and the other data center resources are comprised in different respective data centers.
6 . The computing system of claim 5 , wherein:
the different respective data centers are to communicate with each other via optical communication links and at least one switch.
7 . The computing system of claim 5 , wherein:
the management resources are configurable to dynamically allocate and/or dynamically deallocate, on the as-needed basis, based upon (1) the telemetry-based present resource utilization data, (2) the future resource utilization prediction data, (3) the application programming interface (API) data, (4) the resource utilization trend data, and (5) the quality of service-related data, the compute resources, the storage resources, the accelerator resources, and the local physical rack server resources for use in the execution of the at least one workload.
8 . At least one non-transitory machine-readable storage medium storing instructions for being executed by at least one machine, the at least one machine to be associated with a computing system, the computing system being configurable to be used with local data center resources in association with providing of at least one cloud-based service, the at least one cloud-based service being associated with execution of at least one workload, the local data center resources comprising local physical rack server resources that are to communicate with the computing system via communication links, the computing system comprising other data center resources remote from the local data center resources, the computing system also comprising management resources that comprise circuitry, the instructions, when executed by the at least one machine, resulting in the computing system being configured for performance of operations comprising:
configuring the other data center resources to comprise compute resources, storage resources, and accelerator resources; and
configuring the management resources for use in allocating and/or deallocating the compute resources, the storage resources, the accelerator resources, and/or the local physical rack server resources for use in the execution of the at least one workload;
wherein:
the at least one workload comprises at least one virtual machine workload and/or at least one container workload;
the compute resources, the storage resources, and the accelerator resources are associated with at least one cloud service provider;
the local physical rack server resources are at one or more premises associated with a customer of the at least one cloud service provider;
the management resources are configurable to dynamically allocate and/or dynamically deallocate, on an as-needed basis, based upon (1) telemetry-based present resource utilization data, (2) future resource utilization prediction data, (3) application programming interface (API) data, (4) resource utilization trend data, and (5) quality of service-related data, the compute resources, the storage resources, the accelerator resources, and/or the local physical rack server resources for use in the execution of the at least one workload;
the accelerator resources and the local physical rack server resources are configurable to comprise graphics processing unit circuits for use in the execution, at least in part, of the at least one workload;
the at least one workload is configurable to perform at least one machine learning-related operation;
the management resources are to select, based upon the API data, the compute resources, the storage resources, the accelerator resources, and/or the local physical rack server resources for dynamic allocation and/or dynamic deallocation for use in the execution of the at least one workload;
the API data is configurable to be communicated to the management resources in accordance with multiple different protocols.
9 . The at least one non-transitory machine-readable storage medium of claim 8 , wherein:
the local physical rack server resources are configurable to comprise physical rack server compute resources, physical rack server storage resources, and physical rack server accelerator resources for use in the execution, at least in part, of the at least one workload.
10 . The at least one non-transitory machine-readable storage medium of claim 9 , wherein:
the management resources are to communicate, at least in part, via an API.
11 . The at least one non-transitory machine-readable storage medium of claim 10 , wherein:
the management resources are configurable to allocate and/or deallocate the compute resources, the storage resources, the accelerator resources, and/or the local physical rack server resources, based upon client request data and resource utilization cost data.
12 . The at least one non-transitory machine-readable storage medium of claim 11 , wherein:
the local data center resources and the other data center resources are comprised in different respective data centers.
13 . The at least one non-transitory machine-readable storage medium of claim 12 , wherein:
the different respective data centers are to communicate with each other via optical communication links and at least one switch.
14 . The at least one non-transitory machine-readable storage medium of claim 12 , wherein:
the management resources are configurable to dynamically allocate and/or dynamically deallocate, on the as-needed basis, based upon (1) the telemetry-based present resource utilization data, (2) the future resource utilization prediction data, (3) the application programming interface (API) data, (4) the resource utilization trend data, and (5) the quality of service-related data, the compute resources, the storage resources, the accelerator resources, and the local physical rack server resources for use in the execution of the at least one workload.
15 . A method implemented using a computing system, the computing system being configurable to be used with local data center resources in association with providing of at least one cloud-based service, the at least one cloud-based service being associated with execution of at least one workload, the local data center resources comprising local physical rack server resources that are to communicate with the computing system via communication links, the computing system comprising other data center resources remote from the local data center resources, the computing system also comprising management resources that comprise circuitry, the method comprising:
configuring the other data center resources to comprise compute resources, storage resources, and accelerator resources; and
configuring the management resources for use in allocating and/or deallocating the compute resources, the storage resources, the accelerator resources, and/or the local physical rack server resources for use in the execution of the at least one workload;
wherein:
the at least one workload comprises at least one virtual machine workload and/or at least one container workload;
the compute resources, the storage resources, and the accelerator resources are associated with at least one cloud service provider;
the local physical rack server resources are at one or more premises associated with a customer of the at least one cloud service provider;
the management resources are configurable to dynamically allocate and/or dynamically deallocate, on an as-needed basis, based upon (1) telemetry-based present resource utilization data, (2) future resource utilization prediction data, (3) application programming interface (API) data, (4) resource utilization trend data, and (5) quality of service-related data, the compute resources, the storage resources, the accelerator resources, and/or the local physical rack server resources for use in the execution of the at least one workload;
the accelerator resources and the local physical rack server resources are configurable to comprise graphics processing unit circuits for use in the execution, at least in part, of the at least one workload;
the at least one workload is configurable to perform at least one machine learning-related operation;
the management resources are to select, based upon the API data, the compute resources, the storage resources, the accelerator resources, and/or the local physical rack server resources for dynamic allocation and/or dynamic deallocation for use in the execution of the at least one workload; and
the API data is configurable to be communicated to the management resources in accordance with multiple different protocols.
16 . The method of claim 15 , wherein:
the local physical rack server resources are configurable to comprise physical rack server compute resources, physical rack server storage resources, and physical rack server accelerator resources for use in the execution, at least in part, of the at least one workload.
17 . The method of claim 16 , wherein:
the management resources are to communicate, at least in part, via an API.
18 . The method of claim 17 , wherein:
the management resources are configurable to allocate and/or deallocate the compute resources, the storage resources, the accelerator resources, and/or the local physical rack server resources, based upon client request data and resource utilization cost data.
19 . The method of claim 18 , wherein:
the local data center resources and the other data center resources are comprised in different respective data centers.
20 . The method of claim 19 , wherein:
the different respective data centers are to communicate with each other via optical communication links and at least one switch.
21 . The method of claim 19 , wherein:
the management resources are configurable to dynamically allocate and/or dynamically deallocate, on the as-needed basis, based upon (1) the telemetry-based present resource utilization data, (2) the future resource utilization prediction data, (3) the application programming interface (API) data, (4) the resource utilization trend data, and (5) the quality of service-related data, the compute resources, the storage resources, the accelerator resources, and the local physical rack server resources for use in the execution of the at least one workload.