Compare / Fly.io

Watasu vs Fly.io for managed apps in Europe

Last verified: 6 Nov 2026 Prices and limits as of November 2026 Found an error? info@watasu.io

Run your apps, workers and databases on Watasu with fixed monthly pod prices, no per-seat fees and hosting in Germany. Deploy from Git and get managed backups, private networking and WebRTC support on one platform.

Fly.io gives you Firecracker microVMs called Machines, an API to control them and a private WireGuard network. It also offers Managed Postgres. Watasu packages managed data services and long-running app processes around a Procfile, with normal app bandwidth included and app builds billed by running seconds.

This page compares the two hosting platforms on infrastructure, data location, databases, pricing and developer workflow, and then shows how to move a Fly.io app to Watasu. If you are looking for a Fly.io alternative with managed databases and EU data residency for every component, start with the table.

Start buildingMigration guide

Who should choose which

What Watasu and Fly.io have in common

• Long-running services. Both support WebSockets, TCP services and public UDP, and can run background workers without public ingress.
• Dockerfile or buildpacks. Both support Dockerfile builds and Cloud Native Buildpacks. On Watasu, a root Dockerfile takes precedence.
• Resource billing. Fly.io charges for resources and services used. Watasu draws charges from a prepaid euro balance and offers fixed monthly pod prices.
• CLI and MCP. flyctl and the Watasu CLI cover deploys, config and logs. Both platforms offer an MCP server for AI assistants.
• Grafana metrics. Fly.io offers managed Prometheus and Grafana. Watasu provisions a free Grafana workspace when you attach Logs, Metrics or Traces.

Pricing

Watasu app builds have limits of 16 vCPU and 30 GiB RAM and cost €0.50/hour by actual running seconds, including failed and canceled builds. Queue time and image-only pipeline promotions have no build charge. Before execution, a temporary hold of up to €0.50 funds at most one hour, or a shorter deadline from a smaller available balance; unused funds return after the build stops. Enough available balance for one second is required, with no saved-card requirement or free monthly build allowance. App runtime examples below exclude build charges unless a build row is shown.

Fly.io bills running Machines by the second, with prices varying by size and region. Its Frankfurt pricing table puts a 2 GB shared-cpu-1x at about $14.65 per 30 days and a 2 GB performance-1x at about $38.08. Volumes, outbound data and other services add to the bill. Stopped or suspended Machines still incur root filesystem storage charges.

Watasu has no per-seat fees. Top up a prepaid balance in euros from €5 and pay for the resources you use. Each process picks one of two billing models:

• Standard: a fixed monthly price per replica for sustained load, for example €15 for 1 vCPU and 2 GiB.
• Dynamic: measured CPU and memory (€20 per vCPU-month, €10 per GiB-month) with a hard monthly ceiling per replica, for quieter APIs, staging and review apps.

Normal app bandwidth is included. App builds cost €0.50/hour by running seconds. If the balance runs out, apps pause until you top up.

Example monthly costs for continuously running services, managed PostgreSQL and outbound app traffic:

Key differences

Infrastructure: Machines you operate vs processes you declare

Fly.io apps run on Machines, which are Firecracker microVMs. Choose their size and region, group them in fly.toml and manage their lifecycle through the Machines API. Fly Proxy can automatically stop, suspend and start Machines configured for it. Fly Volumes are local to a host, so applications using them need their own replication strategy.

On Watasu, your app is a set of processes defined in a Procfile, and each long-running process has its own container and replica count:

• web for public HTTP and WebSockets, with managed TLS
• worker for queues, consumers and schedulers, not reachable from the internet
• *-tcp for private TCP and gRPC services, reachable only by apps you allow
• *-rtc for WebRTC and voice, with a dedicated TURN gateway and public UDP
• release for migrations that run before each new version goes live

Each process gets its own size, from a small Hobby pod up to 16 vCPU and 30 GiB of RAM, and its own replica count. Once a process has two or more replicas, Watasu spreads them across separate eligible EU zones at no extra charge.

Where your data lives

Fly.io, Inc. is a U.S. company incorporated in Delaware. You can run Machines in 17 regions, including five in Europe. Managed Postgres is available in a separate set of regions, including Amsterdam, Frankfurt and London. Check the service-specific locations for Sprites and partner services.

The U.S. CLOUD Act can apply even when data is hosted in the EU. Under 18 U.S.C. § 2713, valid U.S. legal process can require a covered provider to disclose data in its possession, custody or control, wherever that data is stored.

Watasu LTD is registered in England and Wales. Watasu hosts apps and add-ons in Germany. Customer data, including images, backups, logs, metrics, traces and object storage, stays in the EU.

Databases and backend services

Fly.io Managed Postgres includes high availability, backups and connection pooling on every plan. Plans range from $38 to $1,922 per month, with storage charged separately at $0.28 per provisioned GB per month. Upstash Redis and Tigris object storage are operated by partners and billed through Fly.io.

Watasu runs its data services itself, next to your apps, under one bill:

Attach a service with one command and Watasu injects the connection variables into your app. PostgreSQL Standard and Premium include scheduled backups. PostgreSQL also supports restoring a dump into a replacement database and promoting it after verification.

Real-time, voice and long-running work

Fly.io supports public UDP on a dedicated IPv4 address. Bind the UDP service to fly-global-services and use the same port inside and outside the Machine. Fly.io also supports WebSockets through its proxy.

On Watasu these are ordinary processes. Web processes hold WebSocket connections, worker processes pull jobs for as long as there are jobs, and *-rtc processes get a TURN gateway, a public UDP port and a stable hostname for each replica. You bring your SFU, such as LiveKit Server, mediasoup or Pion. This makes Watasu a natural home for voice agents that need low latency for European users.

AI sandboxes and agents

Fly.io Sprites provide persistent Linux environments with checkpoints and an HTTPS URL. Billing meters CPU time and memory in use, with separate storage charges. Fly.io also offers a hosted MCP server for Machines, Managed Postgres and Sprites.

Watasu sandboxes run in the EU and support templates, SSH, file and command APIs, pause and resume, checkpoints and outbound network policies. A sandbox configured with 2 vCPU and 1 GiB costs about €0.16 per running hour for compute and memory. Basic has a free base plan and supports up to 20 concurrent sandboxes. Pro is €150 per month and supports up to 100. Runtime and retained storage are billed separately.

Developer workflow

On Fly.io, fly launch detects the project and can generate a Dockerfile and fly.toml. Use fly deploy, fly secrets set, fly scale and fly logs for day-to-day work. Fly.io documents a GitHub Actions workflow for per-PR review apps.

On Watasu, one app holds all of its processes and attached services, and a team can share services across apps. Pipelines move the same tested image from staging to production without a rebuild, and review apps can run automatically when enabled with app.json. Access is set per app or pipeline with four roles: Viewer, Deployer, Operator and Maintainer. Attach Logs, Metrics or Traces to get a free per-app Grafana workspace. Dashboards in your repository can sync on release. You can also connect Claude or another AI assistant over MCP to inspect apps, query logs and manage resources within the connection’s permissions.

When Fly.io is the better fit

Fly.io is a strong choice when geography or control drives the decision:

• Products with users on several continents that benefit from Machines in 17 regions and anycast routing.
• Workloads that need the Machines API, suspend and resume, scale to zero, configurable TCP and UDP services, or a private WireGuard network between regions.
• Agent products built on Sprites, with persistent disks and checkpoints per agent.

If your product is a web app, an API, workers and a database for European users, and you would rather declare processes than operate VMs and volumes, Watasu brings those services together with fixed monthly prices for Standard pods.

Migrate from Fly.io to Watasu

Start with your Dockerfile and map your Fly process groups to a Watasu Procfile. Sign up at dash.watasu.io and verify the app, database restore and networking before switching traffic.

1. Create the app and connect GitHub. Create an app in the dashboard or with the CLI, then connect your repository for automatic deploys, or push directly with git push watasu main.
2. Define your processes. Adapt your Dockerfile as needed and map each fly.toml process group to a Procfile entry. Make sure the web process listens on 0.0.0.0:$PORT. Move scheduled tasks into a worker process.

   web: npm run start
   worker: node worker.js
   release: npx prisma migrate deploy

3. Add your databases. Attach PostgreSQL, Valkey or Object Storage with one command, for example watasu addons:create postgresql --app my-app. Watasu injects the connection variables. Connect to Fly Managed Postgres with fly mpg proxy, then use pg_dump to create a custom-format dump. Follow the Watasu PostgreSQL restore workflow and verify the replacement before promotion. Copy Tigris bucket data to Watasu Object Storage with an S3-compatible client.
4. Copy your environment variables. Retrieve secret values from your secret store and recreate the settings on Watasu. Import a dotenv file with watasu config:set --file .env --app my-app and use --secret for runtime secrets. Keep connection variables managed by their add-ons.
5. Point your domain. Add your custom domain to the app, update the DNS record, and Watasu issues and renews the certificate.
6. Turn on review apps and pipelines. Connect the staging branch, configure review apps with app.json, and promote tested builds to production without rebuilding.

Moving a larger workload, or want to see your stack running on Watasu before you switch? Write to info@watasu.io and we'll set it up with you.

Other comparisons

• Watasu vs Vercel Long-running processes vs serverless functions
• Watasu vs Heroku A familiar Procfile workflow, hosted in Europe
• Watasu vs Railway Fixed monthly pod prices and included app bandwidth
• Watasu vs Render Always-on services, data residency and bandwidth pricing
• All comparisons Watasu against other hosting platforms

Frequently asked questions

Put your next app on Watasu

Deploy from Git and run your apps in Germany with managed data services, fixed monthly Standard pod prices and no per-seat fees.

Start buildingSee pricing