{{ $t('productDocDetail.guideClickSwitch') }}
{{ $t('productDocDetail.know') }}
{{ $t('productDocDetail.dontRemind') }}
6.11.3
{{sendMatomoQuery("Sangfor Cloud Platform (SCP)","Resource Overcommitment Principles")}}

Resource Overcommitment Principles

{{ $t('productDocDetail.updateTime') }}: 2025-12-17

CPU Overcommitment:

The total number of vCPUs for all VMs on a single host should not exceed twice the number of the host's total physical CPU threads.

Example: The server in the diagram has 2 physical CPUs, each with 8 cores and 16 threads, resulting in 32 total physical threads. The recommended total vCPU configuration should not exceed 64 vCPUs.

The maximum number of vCPUs for a single VM can be configured up to the total number of physical threads of the largest CPU host within the cluster.

Example: The maximum allowable vCPU configuration for a VM on the server in the diagram is 32 vCPUs.

Memory Overcommitment:

Memory is a more fixed resource. Memory overcommitment is prohibited in production environments. This means the total configured memory for all VMs on a physical host must not exceed the host's available physical memory.