| 摘要 |
Novel tools and techniques for implementing a virtualization congestion control framework. In one aspect, an orchestrator might be provided within a virtual machine environment context in order to provide two-way communications between the virtual machine (“VM”) and one or more applications running on one or more virtual machines in the VM environment in order to control congestion in hardware resource usage, perhaps using a congestion API. In some embodiments, the two-way communications might include communications from the VM to the applications including maximum hardware resources and current resources, and might further include communications from the applications to the VM including pre-congestion notifications and low-utilization notifications. According to some embodiments, a buffer utilization feedback may be provided between the VM and the applications, said buffer utilization feedback allowing the applications to control pushback mechanisms, said pushback mechanisms including mechanisms for pushing back on or decreasing hardware resource usage. |
| 主权项 |
1. A system for implementing a virtualization congestion control framework, comprising:
a host system connected to a service provider network, said host system comprising one or more processors, one or more hardware components, and a non-transitory computer readable medium having encoded thereon a set of instructions executable by the one or more processors to perform one or more operations, the set of instructions comprising:
instructions for establishing two-way communications between an orchestrator of a virtual machine environment, which is established within the host computer system, and one or more applications running on one or more virtual machines provisioned by a hypervisor separate from the orchestrator in the virtual machine environment, wherein the two-way communications include a system pre-congestion notification from the orchestrator to the one or more applications to begin load shedding prior to reaching a limit on available resources of the host computer system, an application pre-congestion notification from the one or more applications to the orchestrator indicating increase in resource usage of the one or more applications that may result in congestion, or a low utilization notification from the one or more applications to the orchestrator indicating that resources of the one or more applications are being under-utilized;instructions for determining, with the orchestrator, whether mass congestion indicators are present, wherein the mass congestion indicators include indications of resource usage of the host system exceeding a threshold level;instructions for, in response to a determination that the mass congestion indicators are present, sending, with the orchestrator, the system pre-congestion notification to at least a first application of the one or more applications to begin load shedding and to reduce use of resources running on the one or more virtual machines in the virtual machine environment;instructions for, in response to the orchestrator receiving the low utilization notification from at least a second application and the application pre-congestion notification from at least a third application of the one or more applications, sending, with the orchestrator, signals to recover resources from at least the second application, and signals at least the third application to increase resource usage; andinstructions for modifying, with the orchestrator, an allocation of resources of host computer system by decreasing allocation to at least the first application and the second application of the one or more applications and increasing allocation to at least the third application of the one or more applications, based on the two-way communication between the orchestrator of the virtual machine environment and the one or more applications. |
