VM placement groups
Once VMs are created, they are automatically distributed across the Yandex Cloud physical infrastructure. To manage hardware fault tolerance levels, you can combine your VMs into placement groups.
VM placement groups are groups of VMs, each hosted on physical hardware as per a strategy. Yandex Compute Cloud uses spread placement and partition placement strategies.
To learn about the limits on the number of VM placement groups, see Quotas.
Spread placement
Spread placement is a VM placement strategy that guarantees that each VM instance is hosted on a separate server rack. If one rack fails, the others will continue to run normally. Virtual machines within the same group can reside in different availability zones.
By grouping VMs based on the spread placement strategy, you can ensure a high level of fault tolerance and reduce the risk of concurrent failure of VMs hosted on the same server rack. However, the placement requirements are more demanding, so a physical resource shortage is more likely for VMs in a placement group rather than for the same number of ungrouped VMs.
To learn about the limits on the number of VMs in a spread placement group, see VM limits.
You can manage VM placement groups that follow the spread placement strategy using the management console, CLI, Terraform, or API.
Partition placement
Partition placement is a VM placement strategy that guarantees that VMs from different partitions reside on different server racks in a data center of one of the availability zones. If one of the server racks with the group's partitions fails, only one partition will be affected.
A single placement group may contain up to 5 partitions. To learn about the limits on the number of VMs per partition, see Quotas.
You can manage placement groups that follow the partition placement strategy using the CLI and API.