Self-hosting gives you the apps and the whole job of running them. Pilae gives you the same open-source apps, operated under a contract, on your premises or in Pilae Cloud.
- What you build
- Servers, Docker Compose files, a reverse proxy, backups and alerts
- What we sell
- The operated estate, under a contract with an SLA
- Software licence
- The same open-source apps either way
- Who operates it
- Your team, including nights and holidays
- Who operates ours
- Pilae engineers and the Pilae Agent, with your approval on every change
- Where it runs
- Your premises, any of 10+ Pilae Cloud regions (six in Switzerland and the EU), hybrid or air-gapped
- Migration included
- Yes, at a fixed price agreed before it starts
What each one actually covers
Thirteen things an open-source estate needs done. Their column is filled in from docs.docker.com.
| Capability | Doing it yourself | Pilae |
|---|---|---|
| Deploy | ||
| One-click deploy from a curated cataloguePick the app, name the users, and it is sized, provisioned and running. | No | Yes |
| Migration onto the platform, run for youYour data out of the system it is in today, rehearsed before it is cut over. | No | Yes |
| SSO and identity wired into each appKeycloak, Entra ID, Google Workspace or Okta, configured on the app itself. | Partly | Yes |
| Keep it current | ||
| Updates applied for you, not just announcedThe release is read, staged and applied. You are not the one doing it. | No | Yes |
| Per-app CVE and advisory trackingThe advisory, the affected version and the fix version, per app you run. | Partly | Yes |
| Changes land in a window you approvedYour timezone, your hours, agreed before anything is touched. | Partly | Yes |
| Staged and checked before productionThe upgrade runs somewhere safe first, and the checks have to pass. | Partly | Yes |
| Prove it | ||
| Restore drills, not just backupsThe backup is restored on a schedule, because an untested backup is a hope. | Partly | Yes |
| A record of every restoreWho restored what, from which file, into which service, and whether it worked. | No | Yes |
| A change log an auditor can readEvery action, the person who took it, the outcome, inside your perimeter. | Partly | Yes |
| See it | ||
| Monitoring and alerting on the appsNot just the host: the application, its checks, and someone who answers. | Partly | Yes |
| Logs and metrics exported to your stackOpenTelemetry, or straight into the SIEM your security team already runs. | Partly | Yes |
| Run it | ||
| An agent that operates the estate with youAsk for a change in words; it plans it, shows you, and waits for approval. | No | Yes |
Pilae vs Doing it yourself, in words
A tick says whether something is covered. These say how.
| What is being compared | Doing it yourself | Pilae |
|---|---|---|
| What you are buying | Nothing, at first. You pay in engineering time, every month, for as long as the apps run. | An operating commitment. Your contract names the apps, the regions, the maintenance window and the response times. |
| Getting an app running | A server, a Compose file, a reverse proxy, TLS, DNS and a backup job, written and tested by your team. | Pick the app, name the users, and it is sized, provisioned and running behind one hardened gate on port 443. |
| Keeping up with security | Someone subscribes to each project's advisories and reads them. A scanner can flag images; deciding what to do is still yours. | Every app tracked against its upstream advisories, with the CVE, the affected version and the fix version named in a plan you approve. |
| Applying an update | Whoever has time pulls the new image, often straight on production, and hopes the migration script agrees. | Staged first, checks green, a verified backup taken, then applied inside the window you agreed. |
| Proving a restore works | Backups run. Whether one has ever been restored is a question nobody wants to ask in an audit. | Restores drilled on a schedule and recorded, with the file, the target service and the result. |
| Single sign-on | Configured app by app, each with its own OIDC or SAML quirks, by whoever read the docs. | Keycloak or Entra ID wired into every app as part of onboarding, with groups mapped to roles. |
| Who answers the alert at 02:00 | The person on call, if there is a rota. Often the one who set it up. | Pilae engineers, alerted by 60-second probes and machine metrics, under response times written into the contract. |
| When the person who set it up leaves | The servers keep running. Why they were configured that way leaves with the person. | The runbook and configuration are documented and handed to you, and the knowledge sits with a company rather than with one person. |
You ask for the change. The agent does the work.
Pilae Agent runs inside your perimeter and proposes each change as a plan you read before it runs. Nothing lands without your approval, and every action goes into the same change log as ours.
“Write down how our Nextcloud is actually set up”
Reads the running configuration, the compose files and the proxy rules, and commits a runbook to your repository, so it stops living in one head.
“Prove that last night’s backups restore”
Restores each one into an isolated copy, checks the app starts and the data is there, and records the result where an auditor can read it.
Self-hosting is a real option
If you have a platform team, doing it yourself works. Docker Compose, a reverse proxy and a backup script will run most open-source apps well. Your engineers learn the stack, and nothing sits between you and the servers.
That is worth having. It is also a permanent job, and the job grows with every app you add.
What self-hosting asks of your team
Each app has its own release cycle, its own advisories and its own migration scripts. Someone has to read them, test the upgrade and land it without breaking the data. Someone has to restore a backup now and then, to know it works. Someone has to answer the alert on a Sunday.
Often that someone is one person. When that person is busy, updates wait. When that person leaves, the reasons behind the configuration leave too.
What Pilae operates instead
The same open-source apps, run under a contract as managed operations. The Pilae Agent plans each update, waits for your approval, applies it in your window and records it. The control plane shows apps, machines, deployments, backups, logs and the audit trail in one console.
Apps answer only on a private network, behind one hardened gate on port 443. Probes check each app every 60 seconds, and alerts reach Pilae engineers. Your contract specifies a monthly availability commitment with service credits. The details are on the security page.
Moving from self-hosting to managed open source
Onboarding is fixed-price and agreed before it starts. We rebuild your estate on your premises or in Pilae Cloud, move the data, prove the restores and cut over one app at a time. The migration service covers each step. See pricing, or talk to us.
Moving from Doing it yourself
A migration is a service, not a support article. This is the sequence we run, and it is in the fixed price.
Inventory what you run
Every app, version, volume, domain and cron job, including the ones nobody remembers adding.
Choose where it runs
Your premises, any of 10+ Pilae Cloud regions, six of them in Switzerland and the EU, a hybrid of both, or an air-gapped site.
Rebuild under the control plane
Same apps, same versions, with SSO, backups, monitoring and the private network set up before any data moves.
Move data and prove it
Volumes and databases copied across, checks run, and a restore proven from the new backups.
Cut over app by app
One at a time, in windows you pick, with the old servers standing until each app has settled.
Doing it yourself still fits when
- You have a platform team whose job this is, with the time to do it properly.
- Running infrastructure is part of what your organisation wants to be good at.
- The estate is small, the data is replaceable, and an outage on a Sunday hurts nobody.
Everywhere else, Pilae
The moment the estate has to stay patched, stay proven and stay auditable without someone in your building doing it by hand, the comparison stops being close.