Scale-out architecture

One or more frontend gateways (user-facing UI and OPC-UA aggregation) plus a single backend (the database-connected workhorse), networked together through the gateway network, plus a shared SQL database. Every gateway runs the standard edition by default; the network split is on by default because separating the tiers is the whole point of the demo.

What you get

With the default single frontend:

Gateway	Role	HTTP port	Edition
frontend	frontend	9088	standard
backend	backend	9089	standard

Default database	Postgres
Network split	on: frontend and backend networks
Reverse proxy	optional

Every gateway joins both networks, so a frontend can reach the database the backend owns.

Frontend count

--frontends N adds more frontends. A single frontend keeps the bare name frontend; two or more are numbered frontend-1, frontend-2, … with ports stepping up from 9088, and the backend takes the next free port.

ignition-stack create demo --arch scale-out --frontends 2
# frontend-1: 9088   frontend-2: 9089   backend: 9090

Run it

ignition-stack create demo --arch scale-out

Then:

cd demo
docker compose up -d
# frontend: http://localhost:9088   backend: http://localhost:9089

Network split

The split is on by default. Force it off to put every gateway plus the database on a single shared network:

ignition-stack create demo --arch scale-out --no-network-split

The Docker network split is separate from the Gateway Network described below: one isolates container traffic, the other is Ignition's gateway-to-gateway link.

Gateway network

The frontends and the backend are joined over Ignition's Gateway Network, and that link auto-forms on first boot with no UI approval. Each frontend opens a plain (non-SSL, port 8088) outgoing connection to the backend, and every gateway runs an Unrestricted incoming policy so the connection is accepted on sight. The generated POST-SETUP.md carries a verification readout — naming both ends of each link — rather than a manual approval step.

Plain transport is a deliberate demo-only default. For a cross-host or production deployment, switch to SSL on port 8060 with approved certificates, which reintroduces a one-time cert-approval step; the redundancy guide walks through that variant.

Edition choices

--edge-role controls which role (if any) runs Edge:

• --edge-role none: every gateway runs the standard edition (the default).
• --edge-role frontend: the frontends run Edge.

The backend cannot run Edge. The frontends open a Gateway Network link into the backend, and Edge is a leaf edition that can't be an aggregation target, so --edge-role backend is rejected at config time with that guidance. Edge frontends are fine — an edge → standard aggregation link is allowed.

Redundancy

The backend is the single workhorse, so it is the role you pair for failover. Add --redundant backend to expand it into a master/backup pair that boots already paired; see Add a redundant gateway. The replicated frontend tier scales out horizontally instead, so it cannot be paired.

Reverse proxy

--reverse-proxy external joins a Traefik proxy you already run; --reverse-proxy scaffold also lays down the ia-eknorr/traefik-reverse-proxy README under reverse-proxy/ (override the location with --proxy-path). Omit it for plain host-port mapping. See the reverse-proxy guide.

When to use it

Use scale-out to demonstrate a multi-tier architecture: one or more frontends aggregating data and a backend owning the database and history. The gateway-network link between the tiers forms automatically on first boot; see the seeding matrix for what travels in the seed and what is left to the gateway UI.