diff --git a/README.md b/README.md
index e9c60a3..d79da16 100644
--- a/README.md
+++ b/README.md
@@ -87,14 +87,29 @@ An empty disk on each server, for 'local storage' :
 
   - /dev/disk/by-id/nvme-QEMU_NVMe_Ctrl_incus_disk3
 
+### What has to be tweaked to deploy on 'real' hardware : 
+
 One can edit hosts.yaml.example to fit his needs : 
 
-First of all, use uuidgen to generate a unique UUID for your deployment, and remplace it in hosts.yaml : 
+### Global Vars : 
 
   - ceph_fsid: "e2850e1f-7aab-472e-b6b1-824e19a75071"
 
+First of all, use uuidgen to generate a unique UUID for your deployment, and remplace it in hosts.yaml : 
 Must be changed and be unique per deployment.
 
+  -  ceph_rbd_cache / ceph_rbd_cache_max / ceph_rbd_cache_target
+
+RBD Cache settings can be modified here, here is [the documentation.](https://docs.ceph.com/en/reef/rbd/rbd-config-ref/)
+
+  - incus_name:
+
+Must be changed, it's the actual name of the cluster you create.
+
+  - ovn_name:
+
+Can be changed, it's the name of the OVN network deployed across servers.
+
   ### Baremetal vars : 
 
   - ansible_connection: incus
@@ -118,6 +133,10 @@ Depending on your setup, you should consider changing this.
 
 Depending on your setup, could be 'true', and will default to "sudo" privilege escalation.
 
+  - ansible_incus_project: dev-incus-deploy
+
+Must be commented out, if you're deploying on hardware, using SSH connection.
+
   ### incus_init vars :
 
   -  network: LOCAL: parent: enp5s0 / network: UPLINK: parent: enp6s0