DCIM · IPAM · usage billing
Run the estate you rent out.
Kastrum is infrastructure management built for people who sell dedicated servers: what is in the racks, who it belongs to, what it drew last month, and a portal your customers can actually use.
No card. Every feature. Unlimited customers on every plan.
Metering
The part nobody else does properly
Inventory is table stakes. What decides whether you can bill a customer honestly is whether the power and traffic numbers are real — per device, every few minutes, from the hardware itself.
Every outlet, not just the strip
Where the PDU exposes per-outlet metering, each outlet is polled and graphed on its own — and mapped to the server it feeds. That is the difference between “the rack drew 4 kW” and “this customer drew 71 kWh”.
Per-port traffic, counters not guesses
Interface counters are read straight off the switch and turned into in/out throughput per port. The 95th percentile is computed from the samples — which is what an invoice is allowed to be based on.
A month, per customer, with the amount on it
Energy, billable Mbps and volume roll up per tenant per month at your rates. Export it, or push it straight into your accounting as a draft invoice.
And the rest of it
IP space that stays true
IPv4 and IPv6 in a real hierarchy with utilisation, per-customer assignment and a gateway policy you set. Hand a customer a /29 and let them manage it themselves.
Imported from the switch, not typed in
Ports arrive from an SNMP scan and are attributed to servers. Device support comes from a profile pack built out of the LibreNMS definitions — hundreds of switch and PDU families, and nothing to redeploy when you rack a new model.
- You resell dedicated servers and bill for power or traffic.
- You run infrastructure across more than one site and are tired of the spreadsheet.
- You hand customers IPs and need to know which are free, used or reserved.
- You want customers to reboot their own machine without opening a ticket.
One central application holds the data and the credentials. A small stateless agent sits at each site and dials out to it over a single outbound connection — no inbound firewall holes, and nothing sensitive stored at the site. One command installs it.
Secrets are encrypted at rest. Every change is written to an audit log you can read. Your data exports in full, whenever you like.
Try it against your own racks
Thirty days, every feature, no card. Point an agent at one site and see what comes back.