Architecture of a Lightweight Grid Built for Action
Lightweight and powerful are usually treated as opposites. Here is how Feathergrid is built to do serious work without needing a platform team to run it: what is in place today, and what is coming next.
Feathergrid is built to do serious work without a platform team to run it. It's deployed with Terraform and Docker, access is controlled by custom role-based permissions, and new systems connect through APIs. Single-binary cells and ready-made monitoring rules are coming next.
Here's what "lightweight" means in practice.
Why does lightweight matter?
Because the teams that need answers fastest are rarely the ones with a platform engineering department.
A water utility, a regional distributor or a restaurant group shouldn't need a new team just to keep their software running. If the system is heavy to operate, it becomes one more thing someone has to babysit, and the answers arrive late.
How is Feathergrid deployed?
As code. Feathergrid's infrastructure is defined in Terraform and its services ship as Docker containers.
That matters for three reasons:
- Repeatable. The same definition builds every environment, so staging and production match.
- Reviewable. Every change to the infrastructure is a change to code that someone can read and approve.
- Portable. Containers run the same way wherever they're deployed.
It's deployed with Terraform and Docker, access is controlled by custom role-based permissions, and new systems connect through APIs.
Who can see and do what?
Feathergrid uses custom role-based access control. People see only what their role allows, from a single site up to the whole organization. Network plans add an organization-wide policy for who may approve what.
On top of that:
- Routine steps run on their own. You choose which ones.
- Big calls wait for a person.
- One switch stops everything. Every automated action, at once.
How do new systems connect?
Through APIs. Feathergrid's connectors are built to connect with APIs, so your systems, apps and data sources plug in without a long integration project. Your data lakes and warehouses, CRMs and applications stay where they are, and Feathergrid reads from them.
Sensors on-site connect through a gateway at the edge, where correlation and dispatch can also run. See Feathergrid at the Edge.
Where does it run?
The application and your data run on Microsoft Azure in the United States. Keys are held in Azure Key Vault hardware security modules validated to FIPS 140-3 Level 2. AWS holds only the fingerprints that anchor Feathergrid's tamper-evident record, never your data. Enterprise customers can run Feathergrid in their own cloud. If the record and the ledger are ever in conflict, Feathergrid triggers a rollback to the backup database.
More in Feathergrid on Azure and AWS and on the architecture page.
What's coming next?
Two pieces that make Feathergrid lighter still:
- Single-binary cells. Each Feathergrid cell shipped as one self-contained program, simpler to deploy at a site or in your own cloud.
- Ready-made monitoring rules. Prometheus rules you can load into your own monitoring, so your team sees Feathergrid's health alongside everything else.
We'll update this article when they ship.
The short version
Terraform and Docker to deploy it. Custom roles to control it. APIs to connect it. Azure and AWS to keep the record honest. And no platform team required.
See the architecture page or book a scoping call.
Topics
