A unified infrastructure management platform that replaces Proxmox, Netbox, phpIPAM, and MAAS. DCIM, IPAM, hypervisor management, HA clustering, and eBPF networking — all in one binary.
// Provision a new server client := infra.NewClient("https://infra.koder.dev") asset, _ := client.CreateAsset(infra.Asset{ Name: "web-prod-01", Type: infra.Server, RackID: "rack-a1", Position: 12, IP: "10.0.1.50", }) fmt.Printf("Asset %s at U%d\n", asset.Name, asset.Position) // → Asset web-prod-01 at U12
Everything you need to manage physical and virtual infrastructure, from rack to cluster.
Visual rack elevation diagrams with drag-and-drop placement. Track every U-slot, PDU port, and cable path across your datacenter.
Hierarchical IP address management with automatic subnet allocation, VLAN tracking, and next-available-IP queries.
Track servers, switches, firewalls, and peripherals with serial numbers, warranty dates, and lifecycle status.
Auto-generated L2/L3 topology maps from LLDP and ARP discovery. Visualize uplinks, LAGs, and inter-rack connectivity.
Bare-metal provisioning with PXE boot, cloud-init templates, and automated OS deployment — no MAAS required.
Native KVM/QEMU and LXC container management with live migration, snapshots, and resource quotas — no Proxmox needed.
Built-in Raft consensus for multi-node HA. No Corosync, no Pacemaker — just add nodes and the cluster self-configures.
Sub-second failover with hot standby nodes. VMs and containers resume on surviving nodes in under 1 second.
Kernel-level packet processing with eBPF for wire-speed firewalling, load balancing, and traffic shaping — no iptables overhead.
Built-in image registry for OS templates, ISOs, and container images. Ships with Koder Linux ISOs out of the box.
Scan your network and automatically discover servers, switches, and PDUs via SNMP, IPMI, and Redfish — populate your inventory instantly.
First-class integration with koder-mon (monitoring), koder-vault (secrets), koder-grid (compute), koder-net (SDN), and koder-dns.
Query and manage your datacenter rack layout programmatically.
// List assets in rack A1 assets, _ := client.ListAssets(infra.AssetFilter{ RackID: "rack-a1", SortBy: "position", }) for _, a := range assets { fmt.Printf("U%-2d %-20s %s\n", a.Position, a.Name, a.IP) } // U1 core-sw-01 10.0.0.1 // U5 db-prod-01 10.0.1.10 // U12 web-prod-01 10.0.1.50
Automatically allocate the next available IP in any subnet.
// Get next available IP in subnet ip, _ := client.AllocateIP(infra.IPRequest{ Subnet: "10.0.1.0/24", Hostname: "api-prod-03", Description: "Production API server", }) fmt.Printf("Allocated: %s\n", ip.Address) // → Allocated: 10.0.1.51 // List all allocations in subnet allocs, _ := client.ListAllocations("10.0.1.0/24") fmt.Printf("%d addresses used, %d available\n", allocs.Used, allocs.Available) // → 47 addresses used, 207 available
See how Koder Infra stacks up against existing tools.
| Capability | Koder Infra | Proxmox | Netbox | phpIPAM | MAAS |
|---|---|---|---|---|---|
| DCIM / Rack Layout | ✓ | — | ✓ | — | — |
| IPAM / Subnet Management | ✓ | — | ✓ | ✓ | Partial |
| KVM/LXC Hypervisor | ✓ | ✓ | — | — | — |
| PXE Bare-Metal Provisioning | ✓ | — | — | — | ✓ |
| HA Clustering (Raft) | ✓ | Corosync | — | — | Partial |
| Sub-second Failover | ✓ | — | — | — | — |
| eBPF Networking | ✓ | — | — | — | — |
| Auto-Discovery (SNMP/IPMI) | ✓ | — | Scripts | Partial | ✓ |
| Network Topology Maps | ✓ | — | Plugin | — | — |
| Single Binary Deployment | ✓ | — | — | — | — |
| REST + gRPC API | ✓ | REST | REST | REST | REST |
| Go SDK | ✓ | — | Python | PHP | Python |
Six layers, one platform. Every layer is built-in, not bolted on.
Yes. Koder Infra includes a complete KVM/QEMU hypervisor with LXC container support, live migration, snapshots, and resource quotas. It also adds DCIM, IPAM, and bare-metal provisioning that Proxmox lacks.
Koder Infra uses the Raft consensus algorithm for cluster coordination. Raft is simpler, more predictable, and better tested than Corosync/Pacemaker. Hot standby nodes detect failures via heartbeat and resume VMs in under 1 second.
A single node works perfectly for development and small deployments. For production HA, the minimum is 3 nodes (Raft quorum). You can scale out by simply joining new nodes — no manual configuration needed.
eBPF programs run directly in the kernel with zero context switches. This gives wire-speed packet processing for firewalling, NAT, and load balancing — 10x faster than iptables and more flexible than OVS, with no kernel module compilation.
Yes. Koder Infra includes import tools for Netbox (sites, racks, devices, prefixes) and phpIPAM (subnets, IP addresses, VLANs). Run a single command to migrate your existing inventory.
Any Linux distribution via cloud-init or preseed, plus Windows Server via WinPE. The Image Hub ships with Koder Linux, Debian, Ubuntu, and Rocky Linux templates ready to deploy.
One platform. One binary. No more tool sprawl.