7 min readVPS
Most hosting websites are written to leave you with an impression of ownership. Photographs of server halls, a map with pins, a paragraph about "our state-of-the-art facilities". Very few of those companies own the building in the photograph. Most rent space, or rent servers, from a handful of large operators, and the wording is chosen so you do not ask.
We rent too. We think that is the right choice for a company our size, and we would rather explain why than dress it up. It also turns out that whether a host owns or rents tells you much less about your server than the photographs suggest.
What a data center is, and what owning one gets you
A data center is a building engineered around three problems: keeping power on, keeping equipment cool, and keeping out people who should not be there. In practice that means two independent power feeds, batteries that carry the load until the generators start, cooling with spare capacity so one failed unit does not turn into a hot room, smoke detection sensitive enough to catch a fault before there is a flame, access control, and staff on site at three in the morning. Then there is the network: diverse fiber routes into the building, and connections to the exchanges where carriers meet.
Each of those is a specialist discipline with its own engineers, and each of them costs money whether the building holds ten servers or ten thousand. A large operator spreads that cost across thousands of customers. A hosting company that builds its own facility spreads it across its own servers, so either the customers pay for the concrete or the company runs the building on a thinner margin than the specialists manage.
Owning a building also ties you to one place. A company with its own facility in one city has every reason to put every customer there, whether or not that is where the customer's users are.
What renting gets you instead
When the building belongs to an operator whose whole business is running buildings, those costs become a line on an invoice, and we get to make decisions that an owner cannot.
We can put servers where the customers are. Our KVM VPS range is available in Nuremberg, Germany and in Manassas, Virginia, and you choose the region at checkout. The hardware is the same in both: dedicated AMD EPYC cores, DDR5 ECC memory and NVMe storage. For a company with one building of its own, that would be an expensive promise to keep.
We can keep backups in a different building with a different operator. Offsite backups for our services go to Falkenstein, Germany, which is not where any of the servers they protect are. When you are not committed to your own four walls, that separation is easy to arrange, and it is the most useful property a backup can have.
We can move to a newer hardware generation when the operator does, rather than when the last purchase has finished depreciating. And if an operator's service slips, we can leave. An owner never has to face that decision, which is not the advantage it sounds like.
What renting does not change
The operator's responsibility ends at the hardware and the network. Everything above that line is ours, and it is what you are paying a hosting company for.
| Who | What it covers |
|---|---|
| The operator | Power, cooling, fire protection, physical access, the host hardware, the network into and out of the building |
| Red Water | Choosing the hardware, reviewing every order by hand before it is built, the security baseline on a managed server, operating system and platform patching, offsite backups, the SLA, and the engineer who answers your ticket |
On a fully managed VPS we build the server, apply our security baseline before it enters service, keep the operating system and platform patched, and handle the day-to-day administration. On a self-managed VPS you hold root and do that work yourself, and we keep the hardware, network and virtualization layer healthy underneath you. In neither case does the name on the building do any of it.
So the question of who owns the building decides who fixes the chiller. It does not decide who patches your kernel, who notices your disk filling up, or who picks up when something breaks. Those are the questions worth asking a host, and a photograph of a server hall is there to stop you asking them.
A building's tier rating is not a plan
On August 13, 2026, Namecheap lost its website, its shared, VPS and dedicated hosting, EasyWP, DNS management, Private Email and its own support desk when the cooling failed at the RadiusDC facility in Phoenix that hosted its core operations. In their own account, a major storm took out the cooling, temperatures reached critical levels, and the operator instructed them to take services offline to prevent equipment damage. Recovery happened in stages once temperatures were safe. Namecheap describes the facility as a Tier 3 data center designed with a high level of redundancy, and says it had not seen an incident like it in 25 years.
That is not a dig at Namecheap. A capable operator in a properly specified building had a bad day, and the company affected wrote it up in public, which is more than most would do. The lesson is narrower than "pick a better building". A building's rating, whoever owns it, tells you nothing about the two things that decide how bad a day like that is for you.
The first is where your backups are. If they were in the same room as the server, they went offline with it. Ours are in a different city with a different operator.
The second is what happens next. Nobody's VPS fails over to another region on its own, and a host who implies otherwise should be asked to show you the mechanism. Restoring into another region is engineering work. On a managed server, it is work we do with you. Our SLA sets out what is credited when we miss the 99.9% monthly target, and it is written down rather than implied.
Why we name cities and not operators
Our infrastructure page names Nuremberg, Manassas and Falkenstein, gives you an address to ping in each VPS region, and does not name the companies that run the buildings. There are two reasons for that.
We use more than one operator, and for dedicated servers and custom builds we choose per job. A list of names would either go stale or leave someone out.
Naming them would also hand the operator credit for the part of the service that is ours. The European sites we use are run by operators holding ISO 27001, and if a compliance review needs the names, the certificates and the facility specifications, ask and we will send them. That is a better use of the information than a row of logos.
What to check when a host talks about its infrastructure
Ask where, exactly. You want a city and a country, not "world-class facilities", and if they will not say, ask why.
Ask whether you can test it. A ping target for each region costs nothing to publish and tells you more than a map.
Ask where the backups are, how many days are kept, and whether that is written into the terms. The right answer is a different building, and ideally a different operator.
Ask what "managed" covers. Patching, a security baseline, monitoring, support, or only the hypervisor. Get the list.
Ask what the SLA says. A target with credits attached is a commitment. A percentage on a homepage is not.
Our answers to all five are on the infrastructure page, along with the ping targets. If you are choosing a region for a new server, that is the place to start.