Available datacenters, worker nodes and persistent storage flavors

Available regions and data centers

A Managed Kubernetes Service deploys a Kubernetes Master within a given OVHcloud datacenter. The cloud controller then deploys all its worker nodes and persistent volumes in the same region.

Currently, the service is available in US-EAST-VA (Vint Hill, VA) and US-WEST-OR (Hillsboro, OR) regions. All your compute and storage resources will then be spawned in the dedicated region of the associated public cloud project.

Available worker node flavors

When adding a worker node, the Managed Kubernetes Service will deploy a managed Public Cloud instance in the same region/data center. The worker node types correspond to the standard OVHcloud catalog offered in that region and are billed at standard prices.

We currently support several types of instances with guaranteed and constant resources.

• GENERAL PURPOSE (B2-*, B3-*)
• COMPUTE OPTIMIZED (C2-*, C3-*)
• MEMORY OPTIMIZED (R2-*, R3-*)
• GPU (T2-*, L4-*, L40S-*)
• DISCOVERY(D2-*)

Each type (or family) of instance contains different flavors with a unique amount of vCores, RAM, and local storage.

For more information, please refer to the following page: OVHcloud Public Cloud instances.

Note that we are compatible with the standard and flex variations (Flex offers the same disk size for each flavor). Note, however, that Managed Kubernetes Service does not currently offer worker node resizing.

Also, note that many aspects of your Managed Kubernetes Service can be managed via OVHcloud API.

Available persistent Storage Classes

When adding a persistent volume, through Kubernetes API (or kubectl), it will be deployed using Public Cloud additional disks (Cinder Volumes). We support the following storage classes:

• csi-cinder-high-speed-gen2 storage class based on hardware that includes SSD disks with NVMe interfaces. The performance allocation is progressive and linear (30 IOPS allocated per GB and 0.5 MB/s allocated per GB) with a maximum of 20k IOPS and 1GB/s per volume. The IOPS and bandwidth performance will increase as you scale up the storage space.
• csi-cinder-high-speed performance is fixed. You will get up to 3,000 IOPS per volume, regardless of the volume size.
• csi-cinder-classic uses traditional spinning disks (250 IOPS guaranteed, up to 64 MB/s per volume).

These storage classes are based on Cinder, the OpenStack block storage service. The difference between them is the associated physical storage device. They are distributed transparently on three physical local replicas.

High Speed performance is theoretically best for volumes up to 100 GB. Above 100 GB per volume, you will get enhanced performance with a High Speed Gen2 volume. This is detailed in our Persistent Volumes guide.

LUKS encrypted storage classes

OVHcloud Managed Kubernetes also supports LUKS-encrypted block storage volumes using OVHcloud Managed Keys (OMK). The following encrypted storage classes are available in supported regions:

• csi-cinder-high-speed-gen2-luks - Encrypted version of High Speed Gen2 (progressive performance, SSD NVMe)
• csi-cinder-high-speed-luks - Encrypted version of High Speed (fixed 3,000 IOPS)
• csi-cinder-classic-luks - Encrypted version of Classic (spinning disks, 200 IOPS)

For more information:

• Choosing the right Block Storage class

We will support future classes as soon as they are available in OVHcloud Public Cloud.

Go further

For more information and tutorials, please see our other Managed Kubernetes support guides or explore the guides for other OVHcloud products and services.