Operations

Deployment

Reverse proxy, TLS, health checks, and zero-downtime systemd upgrades.

Brume is a fully managed service. There is nothing for you to install, patch, or operate. This page describes the guarantees we hold the service to, so you can plan around them.

What we run

A single Rust gateway binary, supervised by systemd.
A dedicated Redis for rate-limit counters, separate from any customer infrastructure.
Postgres for durable configuration (projects, keys, rules, overrides, quotas, blocklists, webhooks).
A reverse proxy terminating TLS in front of the gateway.

Everything runs on a single VPS in Frankfurt am Main, Germany (Datalix.eu / Equinix FRA). Your data stays in the EU by default.

Endpoints and probes

Endpoint	Purpose
`https://api.brume.run`	The API base URL
`GET /health`	Liveness — the process is up
`GET /readyz`	Readiness — the gateway can serve checks (stores reachable)
`GET /openapi.json`	The OpenAPI document for the current build

Point your load balancer or uptime monitor at /readyz, not /health: readiness reflects whether checks can actually be evaluated.

Upgrades

Deploys are zero-downtime: the new binary starts, passes readiness, and the old process is terminated. There are no long-lived connections to drain — every check is a short HTTP request — so a deploy never interrupts an in-flight evaluation for more than a single retry.

What that means for you

No client-side infrastructure. Your app calls one HTTPS URL.
No Redis to operate. The counters live on our Redis, not yours.
No version skew. The API, the SDK, and the dashboard ship together.

During soft launch we do not offer uptime SLAs, dedicated infrastructure, or private networking. If you need those, talk to us at enterprise@brume.run.

← PreviousArchitecture Next →Security