Up to 1000 Users

Please check technical requirements for basic and cloud-specific requirements.

Kubernetes Node Groups

Configure Kubernetes node groups as below:

Note

You can use your own node group names to align with your internal policies or naming conventions.

AWS

k8s nodegroup	OS/arch	recommendations	autoscaling	labels (key:value)
Default. CPU: 8 Memory: 32 GB	linux/amd64	m6a.2xlarge	35 - 75
GPU Nvidia T4	Ubuntu 22.04 amd64	g4dn.xlarge	3 - 9	type:gpu, gpu.type:T4, gpu.replicas:4, nvidia.com/device-plugin.config:x4-slicing

Due to complication with gpu-operator nvidia driver and linux kernel version, for GPU node group please follow below tested compatible matrix:

release	k8s version	owner account id	ami name
1.9.x	1.31	099720109477 (Canonical)	ubuntu-eks/k8s_1.31/images/hvm-ssd/ubuntu-jammy-22.04-amd64-server-20250121
1.8.x	1.30	099720109477 (Canonical)	ubuntu-eks/k8s_1.30/images/hvm-ssd/ubuntu-jammy-22.04-amd64-server-20240729
1.7.x	1.30	099720109477 (Canonical)	ubuntu-eks/k8s_1.30/images/hvm-ssd/ubuntu-jammy-22.04-amd64-server-20240729

It's important to pin the ami version date because latest version might have newer linux kernel that are not compatible with our gpu-operator nvidia driver

Azure

k8s nodegroup	OS/arch	recommendations	autoscaling	labels (key:value)
Default. CPU: 8 Memory: 32 GB	linux/amd64	Standard_D8s_v5	35 - 75
GPU Nvidia T4	Ubuntu 22.04 amd64	Standard_NC4as_T4_v3	3 - 9	type:gpu, gpu.type:T4, gpu.replicas:4, nvidia.com/device-plugin.config:x4-slicing

GCP

k8s nodegroup	OS/arch	recommendations	autoscaling	labels (key:value)
Default. CPU: 8 Memory: 32 GB	linux/amd64	n2-standard-8	35 - 75
GPU Nvidia T4	Ubuntu 22.04 amd64	n1-standard-4 + T4 GPU	3 - 9	type:gpu, gpu.type:T4, gpu.replicas:4, nvidia.com/device-plugin.config:x4-slicing

MySQL

AWS

Aurora MySQL: db.r6g.xlarge

Azure

General Purpose D4ads_v5: 4 vCores, 16 GiB RAM

GCP

4 vCPUs, 15GB Memory