A VPS (Virtual Private Server) is a virtual machine with an operating system all of your own, on which you have full administrative access. It sits halfway between shared hosting, where everything is managed but hardly configurable, and full cloud platforms, very powerful but more complex. For a website, a moderately sized web application or a few internal services it is often the most balanced choice. The offers look similar, but the real differences hide in the details.
First question: managed or unmanaged
An unmanaged VPS hands you a running machine and nothing more: updates, security, backups and configuration are up to you. It costs less and is flexible, but requires skills and time.
A managed VPS includes a control panel, system updates, sometimes monitoring and help with configuration. It costs more, but frees you from part of the operational work.
The choice depends on an honest question: who will update the system when a critical security fix comes out? If the answer isn't clear, a managed service is probably the better investment.
The resources to evaluate
CPU: shared or dedicated
Most cheap VPSs offer shared vCPUs: physical cores are divided among several customers. This works well for light or variable workloads, but if a neighbor consumes a lot, performance can drop. Dedicated vCPUs guarantee the cores are yours, at a higher price, and are preferable for databases, builds and intensive computation.
Memory
RAM is often the resource that runs out first. A database, a Python or Node application and a web server can require more memory than expected, and when it runs out the system slows down or kills processes. As a starting point, measure your application's consumption in development and leave a margin.
Disk
The type matters, not just the size. NVMe disks are much faster than old SATA SSDs and incomparably faster than mechanical disks, and the difference shows on databases. Also check whether the space is local or on network storage: the former is usually faster, the latter more flexible and sometimes more resilient to failures.
Traffic and bandwidth
Distinguish between connection speed and included traffic. Many providers include a certain number of terabytes per month and charge for the excess; others throttle the speed beyond a threshold. For a site with many media files this can become the main cost.
Where the server is
The datacenter's location affects three aspects.
- Latency: a server close to your users responds faster. For an Italian audience, a datacenter in Italy or central Europe offers low response times.
- Regulation: if you process personal data of European citizens, a provider with datacenters in the Union simplifies meeting the obligations on data transfers.
- Resilience: keeping backups in a location different from the main server protects against failures affecting an entire datacenter.
The features that make the difference
| Feature | Why it matters |
|---|---|
| Automatic backups | Quick recovery after a mistake, though they should be paired with external copies |
| Snapshots | A return point before major updates |
| Network firewall | Filters traffic before it reaches the server |
| IPv6 | Increasingly relevant, and useful for the future |
| Private network | Connects several servers without exposing internal traffic on the Internet |
| API and automation tools | Let you manage the server with infrastructure as code |
| Emergency console | Access to the server even when SSH or the network doesn't work |
The emergency console is an underrated detail: if you get a firewall rule wrong and lock yourself out, it is the only way back in without turning to support.
Price and terms
The monthly price is only part of the cost. Check:
- whether backups, snapshots and extra IP addresses are included or charged;
- whether the promotional price applies only to the first period and what the renewal will be;
- whether billing is hourly, monthly or yearly, and whether you can resize the server without reinstalling;
- what support response times are and in which languages.
Be wary of offers with very generous resources at very low prices: they often indicate overcrowded servers, where the promised performance doesn't materialize at busy times.
Evaluate before committing
Before migrating an important service it pays to run a trial. Many providers bill by the hour or offer a refund period: start the server, measure disk and network performance, try the support with a concrete technical question and check how easy it is to restore a backup. In my work this trial of a couple of days has prevented more than one wrong migration.
Finally, think about the exit. Using standard tools, such as containers or infrastructure as code, and keeping backups in a portable format makes it easy to change provider if one day it becomes necessary.
In short
- A managed VPS costs more but includes updates and support; an unmanaged one requires in-house skills.
- Evaluate shared or dedicated CPU, sufficient memory, NVMe disks and included traffic.
- Choose the datacenter location based on latency, data regulation and resilience.
- Backups, snapshots, a network firewall, a private network and an emergency console make the difference.
- Try the service before migrating and keep the freedom to change provider.
