Skip to main content
Managed Kubernetes node configurations
Last update:

Managed Kubernetes node configurations

In a Managed Kubernetes cluster, you can create node groups with two types of configurations:

If the default Managed Kubernetes configurations are not suitable, you can use fixed cloud server configurations.

You can compare the available configurations in the Comparing node configurations table.

If the available configurations don't suit you, for example you need more vCPU or RAM, create a ticket — we will find configurations with a different number of resources.

Arbitrary configurations

In arbitrary node configurations, you can specify the required number of resources in any ratio.

You can select local disk or one of the four network disk types as the boot disk for nodes.

Fixed configurations with GPU

You can select a fixed configuration with dedicated GPUs and a specified resource ratio. Detailed GPU specifications can be found in the Create a Managed Kubernetes cluster with GPUs instructions.

The configurations correspond to GPU Line cloud servers.

You can select one of four network disk types as the boot disk for nodes.

Comparison of node configurations

If the default Managed Kubernetes configurations are not suitable, you can use fixed cloud server configurations.

Arbitrary configurations
(except ru-3, ru-9,
ru-7, ru-8)
Arbitrary configurations
(except ru-3, ru-9,
ru-7, ru-8)
Fixed configurations with GPU
vCPU cores1-81-324-48

1-8 GPU
RAM1-64 GB*4-256 GB*24-700 GB
Local disk size20-512 GB**20 GB — 1.2 TB**
Network disk size20-512 GB20 GB — 1.2 TB30-512 GB

*If the configuration has more than 8 vCPU cores, you can select RAM with a ratio of at least 1:2. For example, if you choose 10 cores, the RAM should be at least 20GB.

**If the configuration has more than 8 vCPU cores, you can select a local disk size with a ratio of at least 1:32. For example, for 10 cores, the disk size is a minimum of 320 GB.

You can view the availability of configurations in the Managed Kubernetes and GPUs for cloud servers and Managed Kubernetes availability matrix.