How flexibility can get you trapped when using cloud hosting.

Use tools what they made for

To drive 50 km to work every day with a cab is a very expensive solution.

Didier Stickens

Recently there have been some publications of companies that were able to save substantially by moving away from the cloud (see link below). We ask the question: is cloud hosting too expensive? The answer is yes AND no. In this blog, we explain the situation in further detail.

Critical mass

For a start-up or small SME, choosing a cloud provider to host data and applications is an obvious choice. While it is usually more economical to purchase the hardware yourself, it also requires technical skills to manage that infrastructure. In addition, a cloud provider offers a range of attractive benefits such as easy adjustment of capacity to needs, multiple physical locations for servers or backup possible, guaranteed uptime, automatic updates, etc.

However, when the requirements exceed a certain critical mass, it can be quite rewarding to hire in-house technical experts. Indeed, managing large volumes of cloud capacity requires specialized skills and expertise. From a certain volume, it is certainly advisable to study the comparison with the “on premise” (in your own building) or “co-location data center” (your own hardware in a leased location in a data center) alternative.

Flexibility as a pitfall

Let’s take the cab as an example. For those who live in the city and rarely need to travel long distances, the cab is undoubtedly a more economical means of transportation than one’s own car. The cab offers far-reaching flexibility with no driver’s license and no worries about maintenance costs and so on. On the other hand, for those who have to drive 50 km to work every day from the countryside, the cab is a very expensive solution.

We see a similar situation with cloud hosting. Many companies have rightfully stepped into a cab scenario but do not notice however, that in the meantime they also do a lot of long repetitive trips with that cab over the years. Cloud providers are aware of this and do everything they can to retain customers in the long term. In order to do so, they offer all kinds of discounts such as signing up for longer-term capacity on advance payment. This results in less flexibility and unclear billing.

Those considering reducing their cloud bill are advised to hire specialized “cloud consultants”. These specialists will optimize and maintain the cloud. So again, technical experts are needed. There are a lot of companies that make only very limited, if any, use of the various benefits of cloud hosting. For example, internal administrative applications rarely need great flexibility in capacity. In such situation, a lot of savings can be made.

Finally, keep in mind that the following statement is usually true: the longer you are in the cloud, the harder it becomes to move away from it. Also read our blog on vendor lock-in: How to avoid vendor lock-in in the context of software? – Coteng News. By offering more and more so-called “micro-services” – pieces of software that developers no longer have to develop themselves – cloud providers succeed even more in tying your company to them.

What can be saved?

Ahrefs, a large SAAS player with 850 servers, shows in its blog that the company has saved $400 million in 3 years. It’s a rough calculation just based on costs for hardware, rental location and energy. They conclude that their cloud provider is 10 times more expensive than in-house equipment in a shared data center. Because of the specifics of the company, Ahrefs is not fully representative of an average company.

At Coteng, we have been using a hybrid strategy (both cloud and on premise) for years and compare regularly. We take into account a lot of factors such as server hardware, network hardware, UPS, energy, cooling, internet connection, licenses, staff cost. Common costs such as internet connection are spread over the full number of internal servers (in our case 58). We assume a 5-year lifetime for hardware, which in reality turns out to be much longer. With the cloud provider, we use the cheapest price (the 3-year savings plan). Depending on the type of server and the necessary licenses, the difference turns out to be 20 to 30% in the disadvantage of the cloud provider we compared.

Such calculation is not easily transferable to other companies. As soon as the number of servers varies greatly, some fixed costs (for example, labor costs) are less efficiently spread. So the main message is to make the analysis and calculation based on your own situation. Unfortunately, many companies now lack the know-how to analyze this. Why not hire a consultant in that case? It’s important to note that many of them are agents of the big cloud providers. 

Those who neglect to set up a “saving plan” with their cloud provider, even when this is operationally possible, easily lose another 20%. The difference with an in-house purchase scenario then runs as high as 40 or 50%.

When is the cloud the right choice?

However, anyone who would think after all these observations that cloud providers are always an overpriced solution is wrong. For a lot of situations, a cloud provider is the better solution. For example, when flexibility is really needed or when servers on different continents are desired, the cloud provider is the better option. Also, if you want to set up a complex system like a Kafka event streaming platform for the first time, you better start with a ready-made setup from your cloud provider. 

In conclusion, you can see that a certain critical mass is needed to make servers in-house profitable. Anyone who starts blindly deploying everything to a cloud provider and does not regularly evaluate the situation risks paying way too much. So use the right tools for your situation and don’t drive a cab 50 km to work every day.