FOR LEAVING A PAAS

Keep the workflow, change what you are billed for.

Leaving a usage-based platform usually means giving up the part you liked — push, build, live — in exchange for a server you now have to babysit. It does not have to. The workflow is the same here; what changes is that the machine is rented in your name.

Scheduled backups is available on the Pro plan and above. Compare plans →

THE SITUATION

What tends to go wrong.

The invoice moves on its own

Bandwidth, build minutes, function invocations and seats are all counted separately, so a good month for the product is a bad one for the bill.

The data lives somewhere else

The app is on the platform and the database is with a second vendor, each with its own pricing and its own limits on what you can reach.

Going back to servers feels like a downgrade

The alternative on offer is usually a bare VPS and a weekend of nginx, certbot and systemd.

WHAT DEPFLOY DOES

How each of those is handled.

The same push-to-deploy

A push builds a release on your server and the symlink flips when it succeeds. Rollback reuses retained PHP-FPM releases. Node.js, Octane and missing releases rebuild at the selected commit; database migrations are separate.

The database moves with you

PostgreSQL, MySQL, Redis, Valkey, ClickHouse and Meilisearch are provisioned on servers you own, on a private network. One vendor relationship instead of three.

Flat Depfloy subscription

Depfloy does not charge per build, deployment or request. Your provider bills separately for capacity and any metered bandwidth or storage.

Backups you can restore, on your own storage

Six data engines plus project files, on a schedule, to your S3 bucket or R2 or SFTP. No storage tier to outgrow, because we never hold a copy.

Something has to watch it now

The platform's dashboard stays behind, so Depfloy watches from both sides — an HTTP health check that requests the URL on an interval you set, as fast as every five minutes on Starter and every 60 seconds from Pro up, and a heartbeat that a scheduled job or a background worker reports into. Three consecutive failures is what counts as down. The limit to know going in is that checks leave from a single region, so a network problem between that region and your machine reads as an outage until you look.

You keep root

SSH access is yours the whole time, which is the thing a platform cannot give you and the reason the migration is worth doing at all.

The interesting question is not whether a VPS is cheaper than a platform. It is which parts of the platform you were actually paying for — and whether the ones you use survive the move. Push-to-deploy does. The edge network does not.

FAQ

Leaving a PaaS questions

How much of my app has to change? +
Usually little, if it runs as a normal server process. What needs a plan is anything tied to platform-specific primitives — edge middleware, image transformation as a service, a proprietary key-value store.
Can I move one project first? +
Yes, and it is the sensible order. Connect one server, move the least important project, and leave the rest where it is until you are convinced.
What do I lose? +
The edge network and the promise that capacity is someone else's problem. One machine is one machine — Depfloy automates the operations but does not pretend otherwise. That trade is the honest centre of this decision.
Is it actually cheaper? +
It depends on your traffic, and anyone who answers otherwise is guessing. Compare the Depfloy plan plus provider capacity, storage and bandwidth charges with your current bill. The cheaper option depends on your workload and provider pricing.

Other ways people use Depfloy

Try it on one server.

7 days free, no credit card. Connect a machine you already have.