Environment setup

To provision the Azure infrastructure for OpenLM Platform, use Azure Kubernetes Service (AKS) and managed Azure services.

For detailed sizing and service specifications, see Requirements.

Infrastructure components

The Azure deployment uses the following managed services:

Component	Azure service	Purpose
Kubernetes	Azure Kubernetes Service (AKS)	Container orchestration
SQL database	Azure SQL Managed Instance	Identity, operational, and reporting data
Cache	Azure Cache for Redis	Caching and session storage
Persistent storage	Azure Managed Disks	Kafka, MongoDB, and other stateful services (about 400 GB)

Note

Kafka and MongoDB run inside the Kubernetes cluster using Helm charts and Azure Managed Disks for persistence, rather than as Azure managed services.

Network provisioning

Use Azure Container Networking Interface (CNI) networking mode for AKS. The following table lists the network requirements:

Item	Requirement
Virtual network (VNet)	Minimum /22 CIDR block (for example, 10.0.0.0/22)
IP capacity	1,024 addresses, of which about 400 are needed for pods, nodes, and growth
Planned pods	About 143
Planned nodes	6–7

AKS cluster

This section covers the node pools and node labels for the AKS cluster.

Node pools

Create 3 node pools:

Node pool	Purpose	VM size	Count	Max pods per node
System	Kubernetes system pods	4 vCPU, 16 GB RAM (for example, Standard_D4ds_v4)	2	30
Main	Core application workloads	4 vCPU, 16 GB RAM	2	70
Reporting	Reporting workloads	4 vCPU, 16 GB RAM	3	30

Use Kubernetes version 1.32.9 or another AKS-supported version matching your release policy.

Node labels

Apply labels to nodes for workload scheduling. The OpenLM Platform chart expects openlm.com/role=main on the main nodes and openlm.com/role=spark on the reporting nodes:

kubectl label node <system-node>     openlm.com/role=system
kubectl label node <main-node-1>     openlm.com/role=main
kubectl label node <main-node-2>     openlm.com/role=main
kubectl label node <report-node-1>   openlm.com/role=spark
kubectl label node <report-node-2>   openlm.com/role=spark
kubectl label node <report-node-3>   openlm.com/role=spark

Managed services

This section lists the requirements for each managed Azure service.

Azure SQL Managed Instance

The following table lists the SQL Managed Instance requirements:

Setting	Requirement
Tier	General Purpose
Compute	4 vCores minimum
Storage	256 GB minimum
Scaling	Increase based on user volume, reporting load, and data retention

Azure Cache for Redis

The following table lists the Azure Cache for Redis requirements:

Setting	Requirement
Tier	C2
Cache size	2.5 GB

Azure Managed Disks

Managed disks are provisioned automatically by AKS when storage classes are configured. Approximately 400 GB total is needed for Kafka, MongoDB, and other stateful services.

In-cluster infrastructure services

Kafka and MongoDB are not provisioned as Azure managed services on this path. They run inside the AKS cluster, backed by Azure Managed Disks. Create their namespace first by running this command:

kubectl create namespace openlm-infrastructure

Kafka

Deploy Kafka from the Helm chart in the deployment package by running this command:

helm install kafka <kafka-chart>.tgz -f values.yaml -n openlm-infrastructure

Set the storage class and per-broker volume size in values.yaml before installing. Allow roughly 250 GB per broker.

Alternatively, use Confluent Cloud through the Azure Marketplace and point OpenLM Platform at its bootstrap servers instead of installing in-cluster. Azure Event Hubs with the Kafka protocol layer isn't supported.

MongoDB

Deploy MongoDB inside the Kubernetes cluster using the Helm chart provided in the deployment package. Azure Managed Disks provide the underlying persistent storage.

helm install mongodb <mongodb-chart>.tgz -f values.yaml -n openlm-infrastructure

MongoDB Atlas through the Azure Marketplace is a supported managed alternative.

Verify the pods

Run this command:

kubectl get pods -n openlm-infrastructure

Before you continue, check that all pods show Running.

Next steps

After you provision the infrastructure:

1. Configure kubectl to connect to the AKS cluster:

   az aks get-credentials --resource-group <resource-group> --name <cluster-name>

2. Complete the Prerequisites checklist.
3. Proceed to Platform installation.