There are some people who, when viewing our website or hearing about us for the first time, have a hard time believing that we can offer such great servers and support at such a low price. This article sheds some light on the reasons behind the amazing price/performance ratio we offer.
It all starts from the ground up. Not only do we own our own data centers, we also own the land they are built on. Hetzner was founded in 1997, and after more than 10 years of success in the web hosting business, we decided to start building our own data centers. Having spent years setting up and configuring data centers within existing buildings, we knew all about the compromises that entailed. Those years gave us the experience to know exactly what we wanted from a data center.
The first new data center was completed in 2009. Since then, we have operated two large data center parks in Falkenstein/Vogtl. and in Nuremberg. In 2018, we officially opened an additional location Tuusula, Finland, just outside of Helsinki.
We have become experts at building these data centers; instead of outsourcing the work, we staff our own building crew. We have electricians, electrical engineers, technicians for air conditioning and other systems, construction supervisors, architects and many, many more. This gives us complete and total control over what we design and build. We also have our own trucks to transport servers to our different data centers. Simply put, we do as much in-house as we can.
By building our own data centers, we are able to influence every single aspect of them. One great example of this is the energy efficiency of our data centers. Each kWh of electricity which we do not use enables us to offer our products and services at a lower price.
It is our aim to offer servers with an excellent price/performance ratio. We do this using a whole series of measures. First of all, we build the servers ourselves. Most people know that building your own computer is cheaper than buying a pre-built one. The same principle applies to servers.
Building our own servers also means we have full control over what goes into them. We select our server components very carefully, and do so on the basis of four aspects:
- Power consumption
- Quality (robustness)
We do not spend a lot of money on nice 1U 19" rack mount cases. Nobody but our technicians see those cases anyway, and we do not need to spend a lot of money on features we do not need. We use an easy, custom made case, which is specially designed for our airflow requirements and comes with a high efficiency and long lasting power supply. Some of our servers also are case-less, further cutting down on the cost of materials and on energy consumption.
The way we spend our money is based on our many years of experience in the web hosting market. For example, we only use Juniper gear in our backbone; we use fully-redundant boxes, and even then, we use two of them in every location to achieve another level of redundancy. However, we still cannot guarantee a 100% stable and hassle-free network, but we are extremely proud of what we can offer at our prices, and our clients are quite happy with our network. We could spend a lot of money on improving our network by just a tiny bit, but we would rather spend our money more efficiently.
We offer well-designed, reliable, and highly-automated products. This allows us to deploy our servers and other products quickly. For instance, the deployment for one of our dedicated root servers looks approximately like this:
Before the customer makes the order:
- 20 days : We order the components for the server (manual process)
- 15 days: We build the server (manual process)
- 12 days: We mount the server in its rack mounted in the data center (manual process)
- 10 days: The server performs a burn-in test for around two days (automated process)
- 8 days: We check all test results and store them in our hardware database (automated process)
- 8 days: We power off the server (automated process)
- 0:00: We receive a server order from the customer (manual process by the customer)
After the customer makes the order:
- less than 1 minute: We approve the order (This is an automated procedure if there are no special requests; otherwise we manually review the order manually during business hours.)
- 1 minute: System sends WOL signal to the server (automated process)
- 2 minutes: Server boots from the network and starts to perform an image installation (automated process)
- 14 minutes: Some post image installation tasks are performed like setting the network config, date and time, password, etc. (automated process)
- 15 minutes: The server reboots into the newly installed operating system (automated process)
- 16 minutes: We send the server login details to the customer (automated process)
We have a very similar process for handling server cancellations. So both processes (ordering and cancellation) are almost fully automated. This process works fast, is reliable, and is very cost efficient — as long as the customer is an existing customer and they order a server with a standard configuration. With a custom server, we need to build it manually, meaning it can take up to 3 days or more for the server to come online. For new customers, we may need some additional time (up to one business day) to authenticate the account. Hetzner Cloud servers and products are usually ready within seconds.
Our prices are based on the actual cost of the server and the support. We try not to cross subsidize our products or services.
If we consider the typical lifecycle of a server, the highest costs arise from the moment when new hardware is purchased. Also, the depreciation of a server is by far the largest during its first month of operation. Therefore, we charge a setup fee for most of our servers (older servers tend not to have a setup fee). There are a number of competitors who offer servers without a setup fee, and some even offer heavily-reduced monthly prices for the first few months. This pricing strategy is completely opposite to the actual costs involved in running a data center, and must be made up in some other way. We feel that the best price/performance ratio lies in keeping our price structure as close to our internal costs as possible.
Many clients ask us about discounts on the setup fee. Each reduction in the setup fee of a server means we need to set the base (monthly) price of the server high enough so that we do not lose any money on discounted servers. The moment we were to offer discounts would also automatically mean that we would need to ask for higher monthly prices.
These are strategic decisions which make the business process fast and uncomplicated and ultimately also very efficient. We do not need to spend any time haggling or trying to convince clients of our special deals. Clients do not need to wonder if they are getting the best deal; every deal is the best deal. Keeping our monthly prices low also means that the longer a customer uses a server, the more the save.
The information on this page gives customers a glimpse into our pricing structure and shows how we can offer such a great price/performance ratio.
Clients can decide if our concept fits their needs. We know that we are not able to handle 100% of clients' requirements; in fact, we do not even try to do that. We try to handle around 95% of potential clients (the mainstream) and do that in an efficient and reliable way. If you are uncertain about us, simply give us a try. We have monthly contracts, so if you don't like us, it is easy to move on.