Data Plane vs Control Plane
To use Compute, several components will be automatically installed into your cluster. These components are split into two parts.

Data Plane

The "data plane" is the component that actually executes your users' functions, and it is comprised of an auto-scaled group of containers running Atmo. In your cluster, Atmo will be running in a pod called scc-atmo.
The Atmo instances are run in 'headless' mode, which means they are taking instructions from a remote source, the control plane in this case.

Control Plane

The "control plane" is comprised of two components; scc-control-plane and scc-builder. These two components are responsible for administering the cluster of Atmo instances and your users' functions, respectively.
    scc-control-plane acts as a 'brain' for Atmo, and controls things like autoscaling, collecting usage and error metrics, connecting to the Suborbital Telemetry service, and providing administrative APIs. It also manages the WebAssembly modules stored in your storage bucket.
    scc-builder is the component that builds your users' functions and provides the embedded code editor. It can compile various languages to WebAssembly, powers the code editor, and connects to your users' GitHub repos to provide CI/CD functionality.
The control plane connects to, our authentication, billing, metadata, and telemetry service. An environment token is needed for the control plane to operate.

Architecture Diagram

An Architecture diagram can be seen here:
Last modified 16d ago
Export as PDF
Copy link