Mr.PlanB Logo

    Free tool

    Network Topology Planner

    Tell it what you have, say 3 servers and 2 storage boxes, and it draws how to arrange them: which switches, which VLANs for management, VMs, storage, cluster and backup traffic, and how to cable for the redundancy you need. Works for homelabs and enterprise racks.

    Your setup

    Example

    Three node Proxmox VE HA cluster with dual switch fabric, shared NAS and PBS backup

    Three Proxmox nodes connect to a redundant pair of 10G switches so every server, NAS and backup target has two independent paths, and corosync, storage, migration and VM traffic are each separated onto their own VLAN. The firewall sits on a dual uplink to the switch pair and the PBS node gives a dedicated backup target, so no single cable, switch or host failure takes the cluster down.

    EDGENETWORKCOMPUTESTORAGEinternetInternetfw1Firewall Routersw1Core Switch 1sw2Core Switch 2pve1Proxmox Node 1pve2Proxmox Node 2pve3Proxmox Node 3nas1NAS 1nas2NAS 2pbs1Proxmox Backup …mrplanb.com/tools/topology-planner

    Tap a network above to highlight its cables. Hover a device or cable for details.

    Recommendations

    • Use three nodes so corosync keeps quorum when one node fails, and add a QDevice only if you ever run two nodes.
    • Keep corosync, storage and migration on separate VLANs so heavy NFS or iSCSI traffic never delays cluster heartbeats.
    • Set MTU 9000 on the storage, migration and backup VLANs end to end, and verify with a large ping before going live.
    • Bond each server NIC pair with LACP across both switches so a switch reboot causes no outage.
    • Place pbs1 on its own backup VLAN and enable a separate management path to it so backups survive a cluster-wide incident.
    • Enable multipath for iSCSI on NAS devices and configure NFS with failover so VM storage stays reachable during a NAS reboot.

    Watch out for

    • Both switches and the firewall are single logical units if MLAG or the firewall pair is not configured, so test failover before go live.
    • NAS devices may be single controller boxes and become a single point of failure for VM disks unless they have dual controllers.
    • Proxmox HA depends on shared storage, so a NAS outage stops HA recovery for VMs even if the nodes stay healthy.

    Networks

    NetworkVLANSubnetSpeed
    Management Network1010.10.10.0/241G
    VM Network2010.10.20.0/2410G
    Storage Network3010.10.30.0/2410G
    Cluster Corosync Network4010.10.40.0/2410G
    Migration Network5010.10.50.0/2410G
    Backup Network6010.10.60.0/2410G
    WAN and Core Uplink9910.10.99.0/2410G

    The planner sends the form above to our server, where Claude (an AI model) designs the topology. It is a starting point, not a validated design: check port counts, switch features such as MLAG, and vendor support before you buy or cable. 3 plans per visitor per day. Questions? Use the contact page.