JSON is the plan, Stork is extra
You already operate Kea. Stork is another install to see it. The spreadsheet is still the used list. That is still two systems.
Replace
Keep your DNS. Option 6 stays put unless you move it.
kea-dhcp4.conf plus Stork (extra) plus a sheet is still a split brain on a busy Tuesday. Lattice is one catalog: Why NAK is a named Work queue, not a log line, and Discover on one node with Ack on another is the happy path — same pool.
On-prem · Ubuntu tarball · Community: 1,000 managed IPs · No phone-home
Community is the install with no license file. A 30-day Starter trial is in the portal. List prices are on Pricing.
Today
The plan usually lives in a sheet (or phpIPAM). Offers still come from Windows, ISC, or Kea. You keep those stories aligned yourself.
Lattice
Prefixes, pools, reservations, and leases are one catalog. dhcpd serves those rows; the console writes them. There is no second copy to sync. A/PTR follow the grant into the DNS you already run — Ack does not wait on a nameserver.
Lattice does not ingest Kea JSON. Map subnet4, pools, and reservations to prefix, subnet, pool, reservation, then POST /api/v1. kea-admin lease-dump can seed reservations for printers that must keep an IP — it is not a live lease import. Kea HA config does not come with you.
Playbook: Migrate · HA is not Microsoft failover — three-node VIP.
JSON plus Stork plus a sheet is still two systems. Lattice is one catalog, so Work can name the failure.
Paste a MAC suffix → land on the lease.
Work names the queue when a grant fails; occupancy is every remaining lease row as-of a day, CSV — not a Stork lease-dump. A printer reservation you save is the next Offer on every node that shares the database.
| Topic | Kea | Lattice |
|---|---|---|
| Plan and leases | JSON + Stork (extra) + often a sheet | One catalog |
| Scale-out / failover | Kea HA pair | Same pool on every node (Starter: extra dhcpd + three-node VIP). A 30-day Starter trial is in the portal. |
| Find a MAC | Stork / logs | Console search |
| Who held this IP | Stork / lease-dump | Every hold. As-of a day. CSV. |
| DNS | BYO / hooks | A/PTR after Ack into DNS you keep |
| Cutover | Map subnet4/pools/reservations; lease-dump seeds reservations only | Dump → map → API |
You already operate Kea. Stork is another install to see it. The spreadsheet is still the used list. That is still two systems.
Kea HA is a relationship between servers. It is not identical dhcpd on a shared lease store. Discover on A and Ack on B is not the happy path.
kea-admin lease-dump is a seed for reservations if you want printers to keep an IP. It is not a catalog dhcpd must honour tomorrow.
No per-node subnet JSON. A reservation is the next Offer on every node that shares the database.
Search, Addresses, Work. Why NAK is a named queue, not a log line in Stork.
Occupancy is every remaining lease row. As-of a day. CSV. Not a Stork lease-dump.
BIND TSIG or AD GSS-TSIG after Ack. Option 6 does not move unless you move it.
Community is one Ubuntu node. Same binaries. A Starter file adds dhcpd processes on the same database (Discover on one node, Ack on another, same pool) and three-node HA behind a VIP. Community is no license file. A 30-day Starter trial in the portal tests HA.