Watasu
Get started
Features

A platform that does its job.

The infrastructure work nobody enjoys — done well, on by default, and out of your way. Here's what Watasu actually does for you.

Build & deploy

Push to deploy

Add a Git remote, push, get a release. Build output streams straight to your terminal so you watch the deploy happen live.

Cloud Native Buildpacks

No Dockerfile? No problem. Buildpacks detect your stack and assemble a container image automatically — skip the Dockerfile golf.

Bring your own Dockerfile

Need full control? Drop a Dockerfile and Watasu builds it as-is. No platform-specific config to learn, no proprietary base images.

GitHub-native delivery

Auto-deploy on branch push, promote built artifacts between staging and production, spin up a preview app for every pull request.

Run & scale

Process types via Procfile

Watasu reads your Procfile and routes each process by its name suffix. Public web, internal TCP, real-time UDP, workers — one config covers all of them.

Independent scaling

Each process type scales on its own — web=3 worker=1, each on a different pod size. No coupling, no cascading restarts when you tune one knob.

Multi-DC, automatic

Apps scaled to 2+ replicas spread across two German data centers (NBG + FSN) for resilience. You do not configure it; you just scale.

Multiple public web processes

Need admin.example.com on a separate process? Name it admin-web and attach a domain. Run as many *-web entrypoints per app as your architecture wants.

Release process

Add a release: line to your Procfile for migrations or one-time setup. Runs before each rollout — failures abort the deploy so bad migrations never go live.

Microservices & real-time

Private TCP services via *-tcp

Name a process *-tcp for a private TCP endpoint reachable only by apps you explicitly trust. Service-to-service traffic without VPCs, mesh, or mTLS gymnastics.

WebRTC via *-rtc

Name a process *-rtc and Watasu provisions a dedicated TURN gateway, allocates a public UDP port, gives every replica a stable hostname, and injects TURN_* as env vars. ICE done.

Voice agents & real-time AI

Per-process TURN means call-handling backends get production-grade ICE without an external TURN provider in the request path. Latency budget protected.

Domains & networking

Custom domains with managed TLS

Point a CNAME, run watasu domains:wait. Certificates are provisioned and renewed automatically.

Wildcards and apex domains

*.example.com and example.com are both first-class. Bring your DNS provider; we handle the rest.

Data & state

PostgreSQL with followers

Managed Postgres from €3/mo. Standard tiers add async replicas; Premium adds sync-standby HA.

Valkey (Redis-compatible)

Drop-in cache and ephemeral storage. Speaks the Redis protocol so existing libraries work unchanged.

ClickHouse for analytics

Append-heavy event data, dashboards over millions of rows. Multi-shard tiers for serious scale.

Redpanda streaming

Durable, Kafka-compatible event streams. Producer/consumer patterns with replay.

S3-compatible object storage

Scoped credentials, generous SDK env-var compatibility. Geo-replicated tiers available.

Observability

Grafana, automatic

A per-app Grafana provisioned the moment you attach the first observability add-on. Dashboards, Explore, alerts.

Logs, metrics, traces

Three signals, one Grafana. Use any Prometheus-compatible metrics library and OpenTelemetry traces. Free hobby tiers on each.

Platform metrics included

Pod CPU/memory, restarts, OOMKills, add-on health — published without you instrumenting them.

Operations

Backups & restores

Manual + scheduled backups for PostgreSQL, Valkey, ClickHouse. Restore creates a replacement; promote when ready.

Teams & access

Per-team apps and add-ons. Read-only, deploy, operate, admin — granular and per-app where you need it.

EU-hosted, GDPR by default

EU-only data centers in Germany and Finland. No US data plane, no transatlantic data transfers, no compliance gymnastics.

No vendor lock-in

Standard container images, standard service interfaces. Walk away with the same image any time — nothing to migrate off.

Everything you'd build yourself, already built.

Pick a feature, point at the docs, see the exact CLI command for it.