Managed Kubernetes Product Description

Selectel Managed Kubernetes simplifies the process of deploying, scaling, and maintaining Kubernetes container infrastructure. Selectel is responsible for version updates, security, and the health of the Kubernetes Control Plane.

The product supports user types and roles, projects, and project limits and quotas.

Records of operations with Managed Kubernetes clusters are saved in audit logs.

Versions

Managed Kubernetes clusters support versions 1.34.x, 1.35.x, and 1.36.x.

How Managed Kubernetes works

A Managed Kubernetes cluster can run on:

on cloud servers — Cloud Platform resources are used;
or on dedicated servers — dedicated servers, dedicated server networks, and Cloud Platform resources (load balancers, public subnets, public floating IP addresses) are used.

For more information about the resources used, see the Projects instructions.

As the container runtime (CRI), containerd is used. As the network plugin (CNI) in Managed Kubernetes clusters, you can use Calico or Cilium.

You can work with a Managed Kubernetes cluster in the control panel and via the Managed Kubernetes API. You can also work with a cloud server cluster via Terraform.

When choosing clusters on dedicated servers, consider the limitations.

Cluster composition

Managed Kubernetes clusters consist of:

master nodes — contain the cluster management components, Control Plane. The number of master nodes depends on the cluster type. Master nodes are not displayed in the control panel; Selectel is responsible for them;
worker node groups — contain user application containers. Worker nodes are displayed in the control panel; the user is responsible for them.

Cluster types

Selectel provides three types of Managed Kubernetes clusters: fault-tolerant multi-zonal, fault-tolerant, and basic.

You can only select the cluster type when creating a cluster. After the cluster is created, the cluster type cannot be changed.

	Fault-tolerant multi-zonal	Fault-tolerant	Basic
Number of master nodes	3	3	1
Fault tolerance	Master nodes are located in different segments of a multi-AZ pool. Each segment is in a separate data center. If one of the data centers is unavailable, the Control Plane will continue to operate	Master nodes are located: in different segments of a single-AZ pool — if the pool has multiple segments; or in one segment of a single-zonal pool on different hosts — if the pool has only one segment If one of the three master nodes is unavailable, the Control Plane will continue to operate	The master node is located in one segment of a single-AZ pool. If the master node is unavailable, the Control Plane will not work
SLA	✓ (99.98%)	✓ (99.98%)	✗
Functionality	Full functionality available	Full functionality available	Patch version auto-update unavailable
Suitable for	For production environment (production)	For production environment (production)	for development environment (development); testing environment (testing and staging); pet projects

Limits

Cloud server
Dedicated server

Maximum number of fault-tolerant Kubernetes clusters in one pool for one project	10
Maximum number of basic Kubernetes clusters in one pool for one project	10
Maximum number of node groups in one pool for one project	100
Maximum number of nodes in one node group	15
Maximum number of vCPUs per node	32 `*`
Maximum amount of RAM per node	256 `*` GB
Maximum boot disk size per node	1.2 TB
Maximum number of pods per node	100
Maximum number of persistent volumes (PVs) on a single node	256
Minimum size of a single persistent volume (PV)	1 GB

* You can create nodes with more vCPUs and RAM — use fixed cloud server configurations.

Cluster limitations on dedicated servers

Managed Kubernetes clusters on dedicated servers do not support:

use of custom dedicated server configurations;
adding existing dedicated servers to a cluster;
adding multiple node groups when creating a cluster;
adding a node group in locations where more than one VLAN is allocated for a private network;
automation: autoscaling and auto-healing;
attaching a persistent volume (PV) backed by cloud platform network volumes;
using Terraform.

Areas of responsibility

Selectel is responsible for

creation and availability of master nodes;
creation of worker nodes;
upgrading Managed Kubernetes cluster versions;
monitoring master nodes;
providing node autoscaling capability;
providing node auto-healing capability;
data storage security in compliance with Federal Law No. 152-FZ;
integration with Selectel services;
technical support.

Selectel is not responsible for

managing the Managed Kubernetes cluster;
managing worker nodes;
application creation;
initiating scaling and upgrades.

If you need assistance with Managed Kubernetes cluster administration, order managed services.