Bereitstellung auf Ihrer eigenen Infrastruktur.
Selbst-gehostete Container-Plattform auf Basis von Docker Swarm und Caddy. Multi-Node-Clustering, automatisches TLS und persistente Datenbanken ohne Kubernetes-Komplexität. 100% Open Source unter Apache 2.0.
| SERVICE | IMAGE | REPLICAS | PORTS | STATUS |
|---|---|---|---|---|
| caddy-ingress | caddy:2.8.4-alpine | 3/3 (global) | 80:80, 443:443 | healthy |
| upstand-api | upstand/core:0.14.0 | 3/3 (replicated) | 3000/tcp (internal) | healthy |
| postgres-primary | postgres:18-alpine | 1/1 (pinned) | 5432/tcp (internal) | healthy |
| redis-queue | redis:8.0-alpine | 1/1 (replicated) | 6379/tcp (internal) | healthy |
| upgal-agent | upstand/upgal:latest | 1/1 (mcp-daemon) | stdio / unix-socket | idle |
01 — Drei Bereitstellungsformen
Cloud, selbst-gehostet oder beides.
Eine Plattform, drei Betriebsmodelle — wechseln Sie jederzeit frei.
Upstand Cloud
Deployen Sie Web-Apps direkt aus dem Repository. Verwaltete Control Plane, automatische Backups und zentrale Logs.
- Verwaltete Builds und App-Runtimes
- Automatische HTTPS-Domains ohne Konfigurationsaufwand
- Zentrale Audit-Logs und Team-Rechteverwaltung
- Geplante Backups in S3-Objektspeicher
Eigene Server
Betreiben Sie die gesamte Plattform auf Ihren eigenen Maschinen. Jeder Linux-VPS (Hetzner, OVH, AWS) oder Bare-Metal.
- Agentenlose Serververbindung via SSH
- Natives Docker Swarm Multi-Node Clustering
- Keine proprietären Daemons auf Ihren Hosts
- Volle Datensouveränität und Offline-Fähigkeit
Mix & Match
Cloud-Verwaltung verknüpft mit Ihren privaten Servern über ein verschlüsseltes WireGuard-Mesh.
- Produktionsdatenbanken privat, Preview-Zweige in der Cloud
- Workload-Verschiebung ohne Umschreiben von Konfigurationen
- Einheitliches Dashboard und CLI für alle Umgebungen
- Gemeinsame Team-Berechtigungen und Deployment-Historie
Cloud ⇄ Self-hosted. Ein Klick, jederzeit.
Ihre Anwendungen sind standardisierte OCI-Container und Ihre Dienste reguläre Docker-Images. Verschieben Sie Workloads ohne Neubau oder Ausstiegsgebühren.
02 — Unser Anspruch
Wir glauben, dass sich das Internet menschlich anfühlen kann.
Upstand ist ein unabhängiges Produktstudio für präzise Werkzeuge, verlässliche Infrastruktur und ehrliche Software.
Wir bauen, was wir uns selbst wünschen: Open-Source-Tools für Entwickler und fokussierte Plattformen für das Web. Unsere Arbeit verbindet Software, Handwerk und Unabhängigkeit.
Handwerk vor Bequemlichkeit
Langlebige und saubere Primitiven. Keine Dark Patterns, keine aufgeblähte Telemetrie und kein Vendor Lock-in.
Souveränität vor Abhängigkeit
Sie müssen Eigentümer Ihrer Infrastruktur, Daten und Tools bleiben. Offene Standards und native Container-Runtimes.
Klarheit vor übertriebener Komplexität
Pragmatische Softwaretechnik. Docker Swarm statt Kubernetes-Überbau. Konfigurationen, die man in fünf Minuten versteht.
03 — What we're making
A collection of focused things. One Upstand point of view.
Open source tools, creative platforms, and developer infrastructure built with care.
Upstand PaaS
Multi-node clustering, dynamic reverse proxying, persistent databases, and AI-assisted cluster operations with zero vendor lock-in.
OpenRive
High-performance vector animation engine and format converter. Run dynamic interactive motion across web and native targets.
AdsParty
Direct, contextual advertising exchange without behavioral tracking, third-party cookies, or invasive user profiling.
Tongo
Infinite multiplayer canvas for visual planning, system diagrams, and architectural brainstorming with local-first sync.
dgrunner
Secure, self-hosted build runners that spin up isolated microVM sandboxes in under 200ms and tear down cleanly.
lettersbox
Publishing and subscriber platform with cryptographic proof of delivery, plain markdown authoring, and no spy pixels.
04 — Platform capabilities
Everything needed to ship and run containerized workloads.
Engineered on battle-tested Linux foundations: Docker Swarm, Caddy, WireGuard, and PostgreSQL.
Deploy & Release
Deploy multi-container stacks directly from standard Docker Compose files.
Trigger builds on git commits with automated webhook notifications.
Roll back to any previous deployment state with zero downtime.
Build using standard Dockerfile with BuildKit secrets or automatic Nixpacks detection.
Cluster & Orchestration
Native Raft consensus clustering built right into the Docker engine without etcd overhead.
Schedule application replicas across multiple manager and worker servers in different regions.
Gradually roll out new image versions with configurable parallelism and failure rollbacks.
Restart failed tasks and reassign workloads automatically when a node goes offline.
Dynamic Edge & Routing
Zero-downtime routing reloads verified in-process before applying changes.
Free wildcard and sub-domain SSL certificates issued and renewed automatically.
Internal inter-service networking isolated with VXLAN and optional IPsec encryption.
Route multiple applications and microservices behind single domain endpoints.
Persistent Storage
Dedicated PostgreSQL 18, MySQL, Redis, MongoDB, and ClickHouse templates.
Mount host and distributed volume drivers with persistent data guarantees.
Scheduled database dumps directly to AWS S3, Cloudflare R2, MinIO, or Backblaze.
Import and restore database dumps with transactional integrity checks.
Operations & UpGal
Connect any Linux server via SSH keys without installing third-party daemon agents.
Bounded Model Context Protocol (MCP) agent for diagnosing logs and service health.
Stream real-time stdout and stderr logs across all cluster nodes and containers.
Automate cluster management and deployments through scriptable CLI commands and API tokens.
Security & Governance
Authenticate with Google, GitHub, Okta, or generic SAML/OIDC identity providers.
Manage owner, admin, and member permissions with scoped resource boundaries.
Generate granular deployment tokens with specific project read and write permissions.
Track every deployment, rollback, node join, and configuration change.
05 — Architecture
Engineered for operational durability.
A layered topology ensuring control plane isolation, encrypted mesh communication, and zero-downtime edge routing.
Requests arrive via HTTPS to the edge proxy, while developers connect via the web console, CLI, or UpGal MCP interface.
The Hono and tRPC API coordinates state in PostgreSQL 18, utilizing BullMQ queues for asynchronous worker execution.
Manager nodes maintain Raft consensus and schedule container tasks across isolated overlay networks with health monitors.
Caddy validates routing configurations via dry-run before applying hot updates, guaranteeing zero dropped connections.
06 — Compose to Swarm
Standard Docker Compose in. Production Swarm Stack out.
Upstand translates standard Compose files into resilient multi-node Swarm definitions with health checks and rolling update policies.
Zero-friction translation with safety checks
Standard Docker Compose files are built for single-machine development. Upstand parses them and generates resilient, production-ready Swarm stacks without changing your team's existing workflow.
- Automatic replica scaling and placement constraints
- Rolling update configurations with configurable parallelism
- Automatic restart policies on worker node failure
- Zero manual Kubernetes YAML manifests required
version: "3.8"
services:
web:
image: ghcr.io/org/frontend:latest
ports:
- "80:80"
restart: always
api:
image: ghcr.io/org/backend:latest
ports:
- "4000:4000"
restart: always
db:
image: postgres:18-alpine
volumes:
- pgdata:/var/lib/postgresql/data
volumes:
pgdata:07 — In good company
Partners, collaborators, and friends of the studio.
We build with teams who hold the same conviction for craft, performance, and independent software.
Apex Team
High-performance systems engineering team collaborating on distributed systems reliability, edge networking, and container runtimes.
Mansora
Product design consultancy focused on digital craft, human-centered developer interfaces, and typography-first web architecture.
“Good tools are built with good people. We're proud to collaborate with organizations pushing the web forward.”
08 — Technical comparison
Where Upstand is genuinely different.
A factual comparison of architecture, orchestration models, and resource footprints.
| Capability | Upstand | Coolify | Portainer | Dokploy | Kubernetes |
|---|---|---|---|---|---|
| Orchestration Engine | Docker Swarm (Native) | Single Docker / Swarm partial | Docker / K8s agent | Single Docker / Traefik | Kubelet / control plane |
| Multi-Node Clustering | Built-in (Raft consensus) | Experimental / Single host | Requires agent per node | Single node primarily | Built-in (High complexity) |
| Reverse Proxy Architecture | Atomic Caddy (0ms hot reload) | Traefik | Edge agent / manual | Traefik | Ingress-NGINX / Envoy |
| Server Overhead Footprint | ~180MB RAM | ~500MB RAM | ~150MB RAM | ~350MB RAM | >2GB RAM control plane |
| Remote Node Management | Agentless over SSH | SSH execution | Proprietary agent daemon | SSH execution | Kubelet agent per node |
| AI Operational Assistant | UpGal (MCP Bounded ToolLoop) | None | None | None | Third-party add-ons |
| Open Source License | Apache 2.0 (100% Free) | Apache 2.0 | Open Source / Business tiers | Apache 2.0 | Apache 2.0 |
09 — FAQ
Frequently asked questions.
Clear answers to technical and operational questions about Upstand.
10 — Bring us an idea
Start a conversation with our engineering team.
Whether you are planning a cluster migration, building an ecosystem integration, or looking to collaborate on software, we'd like to hear from you.
Open Source Pledge
Yours to run, fork, and ship.
Deploy Upstand on your own infrastructure today. Free, open source, and self-reliant forever under Apache 2.0.
Apache License 2.0 · No telemetry tracking · 100% self-hostable