If the Kubernetes setup does not need any PV/PVC or other storage connections it is possible to reduce AKS base resource usage by disabling parts of Azure CSI handling. The default CSI pods create base resource usage on every pod. Especially when using small AKS clusters and/or nodes e.g. for test environments the wasted resources might be significant overhead.

Example:
kube-system csi-azuredisk-node-vf4js 30m (1%) 0 (0%) 60Mi (0%) 2800Mi (38%)
kube-system csi-azurefile-node-nvt8d 30m (1%) 0 (0%) 60Mi (0%) 600Mi (8%)

Terraform already supports CSI control flags via storage_profile similar to the already provided network_profile. The current defaults are these:

storage_profile {
blob_driver_enabled = false
disk_driver_enabled = true
file_driver_enabled = true
snapshot_controller_enabled = true
}

The ability to control those flags via normal module configuration would help to reduce resource footprints