This overview traces a file request through SteadyLink's edge, API, database, and object storage. It also explains where access checks, caching, and background jobs fit into that path.
Why this architecture
A public asset URL identifies the file rather than a storage location. That keeps embeds intact when a new revision is published. The edge caches responses, the API streams cache misses from object storage, and short-lived tokens control access to private files without exposing storage URLs.
From request to first byte
When a user opens a canonical URL like /a/{id}, the edge looks for a cached copy. On a hit, the response is immediate. On a miss, the API verifies access and streams bytes directly from object storage. Strong validators (ETag) and cache headers allow the edge to warm quickly and serve repeat traffic close to users.
This path is simple by design: stable links, a small number of hops, and no vendor URLs in the wild.
Uploads without bottlenecks
Uploads start at the API, which issues a short‑lived presigned URL. The browser then uploads bytes directly to storage. Afterward, a commit call records the new version in the database and attaches metadata. When common variants are helpful, such as web‑ready sizes, the commit enqueues background work so the UI stays responsive.
Direct‑to‑storage means fewer retries, predictable throughput, and clear separation between metadata and bytes.
Background work, kept off the critical path
A separate worker precomputes common transforms, sweeps temporary uploads, and purges old entries. None of that work needs to delay a file request.
Privacy and access control
SteadyLink supports both public and private delivery. Private assets are served via signed access tokens and organization scoping. Sensitive actions are auditable so teams can review activity and track usage without exposing global data. Raw storage URLs are not surfaced.
Reliability and performance posture
Edge caching near users, efficient streaming from storage, and versioned URLs for immutable assets keep the system predictable. Database indices support quick lookups for assets and versions, while canonical URL design avoids churn in downstream systems and embeds.
Summary
The delivery URL stays separate from the stored revision. Direct uploads keep file bytes off the API path, edge caching handles repeat requests, and audit records capture sensitive changes.