公开(公告)号：US11966765B2

公开(公告)日：2024-04-23

申请号：US17015377

申请日：2020-09-09

Applicant: NVIDIA Corporation

Inventor： Santosh Kumar Ravindranath Shukla ,  Andrew Currid ,  Chenghuan Jia ,  Arpit R. Jain ,  Shounak Santosh Deshpande

IPC: G06F9/455 ,  G06N3/04 ,  G06N3/08 ,  A63F13/30

CPC classification number: G06F9/45558 ,  G06N3/04 ,  G06N3/08 ,  A63F13/30 ,  G06F2009/45583

Abstract: Systems and methods are disclosed for throttling memory bandwidth accessed by virtual machines (VMs). A technique for dynamically throttling the virtual computer processing units (vCPUs) assigned to a VM (tenant) controls the memory access rate of the VM. When the memory is shared by multiple VMs in a cloud-computing environment, one VM increasing its memory access rate may cause another VM to suffer memory access starvation. This behavior violates the principle of VM isolation in cloud computing. In contrast to conventional systems, a software solution for dynamically throttling the vCPUs may be implemented within a hypervisor and is therefore portable across CPU families and doesn't require specialized server-class CPU capabilities or limit the system configuration.

公开(公告)号：US20220075638A1

公开(公告)日：2022-03-10

申请号：US17015377

申请日：2020-09-09

Applicant: NVIDIA Corporation

Inventor： Santosh Kumar Ravindranath Shukla ,  Andrew Currid ,  Chenghuan Jia ,  Arpit R. Jain ,  Shounak Santosh Deshpande

IPC: G06F9/455 ,  G06N3/04 ,  G06N3/08

Abstract: Systems and methods are disclosed for throttling memory bandwidth accessed by virtual machines (VMs). A technique for dynamically throttling the virtual computer processing units (vCPUs) assigned to a VM (tenant) controls the memory access rate of the VM. When the memory is shared by multiple VMs in a cloud-computing environment, one VM increasing its memory access rate may cause another VM to suffer memory access starvation. This behavior violates the principle of VM isolation in cloud computing. In contrast to conventional systems, a software solution for dynamically throttling the vCPUs may be implemented within a hypervisor and is therefore portable across CPU families and doesn't require specialized server-class CPU capabilities or limit the system configuration.