Skip to content

Latest commit

 

History

History
128 lines (111 loc) · 5.62 KB

Home.md

File metadata and controls

128 lines (111 loc) · 5.62 KB
sidebar_position 1
slug /
title About Grid
sidebar_label About Grid
hide_title true
Grid
Open source · Self-hosted

About Grid

Grid is a self-hosted control plane. You describe what should exist as JSON in Git — a project, a cloud, an environment, and a unit. Grid syncs that tree and runs plan, apply, and destroy the same way on every cloud. Each cloud has its own catalog of what Grid can build there.

Control loop: author JSON, sync, plan/apply by engine, state and monitor

Author → Sync → Plan/Apply (by engine) → State & monitor. Everything else wraps this loop.

What Grid is for

Organizations use Grid when they want:

  1. Intent you own — the JSON lives in a Git repository you control.
  2. One release path — plan, apply, and destroy, with audit and live logs, for every cloud and every engine.
  3. A catalog per cloud — the path shape is the same; the types Grid builds are that cloud’s catalog. See provider catalogs.
  4. An exit ramp — generated Terraform and your Git history stay usable with ordinary tools. See Exit ramp.

Engines

Engine Operates
Terraform Cloud resources, managed clusters, node pools, and other module-bank subtypes
Kubernetes workloads In-cluster apps via GitOps (Argo CD / OpenShift GitOps) — Kubernetes workloads

Managed clusters are Terraform units. Workloads inside those clusters are the separate kubernetes engine. Details live in the sidebar under Concepts.

What Grid is not

Not this Because
A hosted SaaS You run grid-core and grid-ui on your own VM or Compose host
A replacement for Terraform or Argo Grid runs them. Terraform units become HCL. Kubernetes units become cluster GitOps
Automatic destroy when JSON is deleted A missing file is marked stale. Destroy is a separate release

How it works

Your Git repo (intent JSON)
        │
        ▼
  GitOps Sync (grid-core)
        │  upsert units · mark missing JSON as stale
        ▼
  Console / API / CLI
        │  plan · apply · destroy (releases)
        ▼
  Engine:
    · terraform   → generate + module bank → remote state + archive/
    · kubernetes  → cluster GitOps (Argo / OpenShift)
        │
        ▼
  Optional host Prometheus / Alertmanager

That loop is the product. RBAC, approvals, Sources, Monitoring, and Topology wrap it. Use the sidebar for layout, unit JSON, clouds, install, and the rest — this page is the orientation.

Who does what

Actor Responsibility
Operators / platform Install Core + UI, remote state, module bank pin, Git remote, auth bootstrap, and cluster GitOps connectivity for workload units
App / infra authors Own the intent repository; open PRs for JSON (and workload) changes
Approvers Gate apply/destroy where environment policy requires it
CLI users Same generate/plan/deploy path on a local machine with cloud credentials (Grid AuthZ is not applied to raw CLI)

Where to go next