EmberSecurity
EmberSecurity watches every EmberNet edge box for the things that matter on a plant floor: a device that should not be on the network, a write to a PLC that nobody planned, a shell where no shell belongs, a file that changed. What it sees comes to one place, and you read it in the EmberSecurity area of EmberCORE.
EmberSecurity watches and reports. It does not change a PLC, and it does not sit in the path of a control loop. The part that decides what traffic is allowed is Plasma, and EmberSecurity reports what Plasma refused.
Four containers on every box
Every edge box runs four EmberSecurity containers. Each has one job and reports on its own.
| Container | What it watches | What it tells you |
|---|---|---|
| Network watch | The PLC leg and the plant network | Traffic that matches a known attack pattern, and traffic that should not exist on that leg |
| OT protocol sensor | The industrial protocols on the wire | Who talked to which PLC, what they asked it to do, writes to it, and every device it has seen |
| Runtime guard | The box itself | Shells started inside containers, and unexpected outbound connections from a process |
| Integrity and posture | Files and configuration on the box | Changes to watched files |
The first three send a heartbeat, so a container that has stopped is shown as stopped. A silent sensor is never mistaken for a quiet network.
On a box where Plasma's edge firewall sits in line, the sensors keep watching the cable through the firewall, including after a failover to the host.
How what it sees reaches you
All of it leaves the box over Flux on TLS port 443, the same outbound connection the box already uses. EmberSecurity opens no other port on the box and needs no other path off the site, so a plant that allows only outbound 443 needs no firewall change to get security reporting.
What travels is events: alerts, protocol activity, and file changes. The sensors' rules are built into each release, so a new rule set arrives as a normal update and the box needs no extra internet access to get it.
On the platform side, events from every box land in EmberSecurity's central store. You never connect to it directly. EmberCORE reads it for you, and only after it has checked who you are.
How tenants stay separate
Every box reports under its tenant and its own box identity, taken from the same provisioning record the rest of the platform uses. EmberCORE applies your tenant's filter on the server for every request, before it asks the central store anything. A tenant user sees only their own tenant's boxes and events. Fireball staff see every tenant. A request for another tenant's alert or box is refused.
This is the same tenant segregation the rest of EmberNet runs on. See Multi-Tenancy.
What you get from it
Security reporting that follows the box instead of the network. A new device on a PLC leg shows up as a first-seen device with its address. A write to a PLC shows up with who sent it and whether it fell inside your maintenance window. A box that stops reporting shows up as silent. The plant network map places each box's devices by Purdue level, and every action taken on an alert is recorded in the audit trail.
Next steps
- The EmberSecurity area: every view, and how to work an alert
- Plasma edge lockdown: what the box allows on the wire
- Zero-Trust Networking: the Flux model the reporting rides on