A managed VPS is a virtual server where the provider takes over part of the management. Which part is the problem: there is no shared definition, and "managed" moves a layer up or down depending on who sells it. At many panel-based hosts it stops at the operating system and the control panel, the host keeps root, and your application is explicitly out of scope. Sometimes software such as Docker is not allowed either. That is a perfectly good product for someone hosting a few websites, but it is not what a team with its own deployments means by "management". You only see the difference when you ask, layer by layer, who does what. For a look at the bottom layer of such a platform, see the architecture behind PlanetNode.
Four layers, one question each
Under every management label sits the same stack. Walk it from the bottom up and ask at each layer: who is responsible here, and what happens when it breaks?
- Infrastructure — hardware, hypervisor, storage and network. The provider always does this, unmanaged included. The questions are where it runs, on whose network, and who you call when a node fails.
- Operating system — the baseline: SSH, firewall, users, logging, and above all patching. Ask how quickly critical updates land, whether reboots are scheduled, and whether the configuration is written down anywhere or only exists on the server.
- Stack — web server, database, runtimes, containers. This is where definitions diverge most. Ask which components are covered, and whether you can install something yourself without voiding the management.
- Application — your own code and its dependencies. Your responsibility almost everywhere. The question is what happens when the provider's update breaks it: who rolls back, and who finds out why?
Two questions cut across all layers. Who has root? And what do you take with you when you leave? At a panel host your configuration often lives in the panel, so leaving is a migration. If it lives as code in a repository, you take it with you.
Where the checklist usually fails
Three items that look fine in a quote and disappoint in practice:
- Backups without a restore test — a backup on the same storage as the server, or one that has never been restored, is an assumption. Ask where it is stored, whether it is encrypted, and how a restore actually works.
- Monitoring without content — "we monitor your server" can mean a ping. Ask what is measured: disk space, certificates, the HTTP response of your service, the protocols you rely on.
- Patching without rebooting — updating packages is not the same as running the new code. The kernel and long-running processes keep the old version until a restart.
Our three levels, per layer
We defined our levels by layer on purpose, so you do not have to guess where "managed" ends:
- Unmanaged — we deliver and monitor the infrastructure on our own network; you run the OS and everything above it, with an engineer on the line when you get stuck.
- Managed — we run the OS and stack as code. Critical security updates within 7 days at the latest, high within 30; that is an upper bound, in practice we are faster. Monitoring with Prometheus and Alertmanager, with HTTP and protocol checks, and off-site, client-side encrypted backups at another location. If one of our updates breaks your application, we roll it back; the application itself stays yours.
- Managed+ — everything in Managed, plus the application layer. We work in your repo and on your automation model, bring your application's metrics into monitoring and act on them, and debug and develop alongside you.
You keep root on all three. The only exception is when you choose software we developed ourselves; even then we help you leave. A restore test belongs to management: a backup only counts once you know you can restore it.
Managed starts at EUR 125 per server per month, excluding VAT and excluding the server. That is a starting price: after a conversation your setup may turn out cheaper or more expensive. The advice is free of obligation, but we make sure you buy what fits.
Still on VMware and planning to switch platform first? Read what a Proxmox migration really costs.
Management you can check
Want to know at which layer your current management stops, or looking for management where you keep root and your configuration? See managed servers. If you combine it with hardware in our data centres, colocation fits underneath, and for the application layer there is software development. Or get in touch and we will walk through the checklist for your servers.