Cluster Requirements

Subsalt provides Terraform templates for each of the major Kubernetes cloud providers to make it easy to set up compatible clusters. The Terraform templates are accessible through Subsalt's CLI tool.

Cluster resource requirements

• Subsalt runs on Kubernetes v1.32+, and can be deployed in multiple configurations depending on your needs. Multiple components can be deployed in a single cluster, or across multiple clusters.

  • We follow Kubernetes release support lifecycle for the versions of Kubernetes that we aim to support.

• Subsalt must have cluster-level permission to add operators at installation time.

• Below are tables that note the minimum requirements for node pools configured in each cloud environment.

Microsoft Azure AKS

Name	Node count	Recommended instance type	Taints	Labels
common	2 (fixed)	Standard_E16s_v3		subsalt.io/node-purpose=common
serving	0-1 (auto-scaling)	Standard_NV12ads_A10_v5	nvidia.com/gpu=present:NoSchedule	subsalt.io/node-purpose=serving, subsalt.io/has-gpu=true
pipelines_cpu	0 - 4 (auto-scaling)	Standard_E16s_v3	subsalt.io/node-purpose=pipelines:NoSchedule	subsalt.io/node-purpose=pipelines, subsalt.io/has-gpu=false
pipelines_gpu	0 - 3 (auto-scaling)	Standard_NC8as_T4_v3	nvidia.com/gpu=present:NoSchedule	subsalt.io/node-purpose=pipelines, subsalt.io/has-gpu=true

Amazon Web Services (AWS) EKS

Name	Node count	Recommended instance type	Taints	Labels
common	2 (fixed)	r6a.4xlarge		subsalt.io/node-purpose=common
serving	0-1 (auto-scaling)	g5.xlarge	nvidia.com/gpu=present:NoSchedule	subsalt.io/node-purpose=serving, subsalt.io/has-gpu=true
pipelines_cpu	0 - 4 (auto-scaling)	r6a.4xlarge	subsalt.io/node-purpose=pipelines:NoSchedule	subsalt.io/node-purpose=pipelines, subsalt.io/has-gpu=false
pipelines_gpu	0 - 3 (auto-scaling)	g5.4xlarge	nvidia.com/gpu=present:NoSchedule	subsalt.io/node-purpose=pipelines, subsalt.io/has-gpu=true

Assuming you're using EC2 autoscaling groups for cluster autoscaling the following tags will also need to be set (assumes recommended instance types).

CPU and Memory values should be set to between 85-90% of the actual instance type's resource values to account for Kubernetes system pods.

Serving node pool

Key	Value
k8s.io/cluster-autoscaler/node-template/resources/cpu	3
k8s.io/cluster-autoscaler/node-template/resources/memory	14G
k8s.io/cluster-autoscaler/node-template/resources/nvidia.com/gpu	1
k8s.io/cluster-autoscaler/node-template/taint/nvidia.com/gpu	present:NoSchedule
k8s.io/cluster-autoscaler/node-template/label/subsalt.io/has-gpu	true

Pipelines (CPU) node pool

Key	Value
k8s.io/cluster-autoscaler/node-template/resources/cpu	14
k8s.io/cluster-autoscaler/node-template/resources/memory	116G
k8s.io/cluster-autoscaler/node-template/taint/subsalt.io/node-purpose	pipelines:NoSchedule
k8s.io/cluster-autoscaler/node-template/label/subsalt.io/node-purpose	pipelines
k8s.io/cluster-autoscaler/node-template/label/subsalt.io/has-gpu	false

Pipelines (GPU) node pool

Key	Value
k8s.io/cluster-autoscaler/node-template/resources/cpu	14
k8s.io/cluster-autoscaler/node-template/resources/memory	56G
k8s.io/cluster-autoscaler/node-template/resources/nvidia.com/gpu	1
k8s.io/cluster-autoscaler/node-template/taint/nvidia.com/gpu	present:NoSchedule
k8s.io/cluster-autoscaler/node-template/label/subsalt.io/node-purpose	pipelines
k8s.io/cluster-autoscaler/node-template/label/subsalt.io/has-gpu	true

Networking

Ingress

Subsalt requires that the cluster has an Ingress controller support for web access.

There are two hosts to configure, one for the web portal and one for authentication management. They should take the form portal.subsalt.acme.com and auth.subsalt.acme.com.

A Record's for each host should be configured in your DNS provider to point at the ingress IP address.

A certificate (TLS/SSL) should be added to the Ingress resource that accounts for both hosts. The Subsalt Helm Chart comes with cert-manager by default which can be configured to provision the certificates and their secrets automatically.

Load balancer

Subsalt requires Service (Load Balancer) support for the query endpoint to serve synthetic data.

An A Record should be configured in your DNS provider to point at this load balancer's IP address.

Other

If your team uses another Kubernetes provider (IBM, DigitalOcean, etc), please reach out and we can find a way to support your deployment.