Is the modular data center ideal in all cases?

The pursuit of greater data center modularity is not new, but it appears to be accelerating today, primarily to streamline investment and operating costs for created or available space. Thus, beyond the initial investment amount to build a data center, companies now anticipate the costs inherent in their operation. And this for several reasons. With rising energy costs, they are fully aware that an oversized data center generates higher costs in the long run. At the same time, due to economic conditions, the rapid evolution of technologies offered by IT manufacturers, and new solutions brought to market by software vendors, some companies have less visibility into the evolution of their business: capacity increases become more difficult to anticipate, both in volume and time.

When they do not decide to have their information system hosted by a third party, with pay-per-use billing, companies are increasingly receptive to the idea of having data centers capable of adapting to the evolution of their needs, progressively.

Several solutions exist to meet this demand. The first is the concrete data center designed in a modular manner. Unlike a conventional data center, it consists of several cells independent of one another. The entire structural work, as well as part of the finishing work (particularly the technical ducts intended to accommodate the various utilities) are completed during construction. Each cell is equipped (electricity, cooling, cabling…) and put into production when the previous one reaches a certain load rate, without any impact on the premises already in operation.

Beyond the fact that this design offers better efficiency at lower energy cost, it is recommended for companies that have strong requirements in terms of security and quality of service: each cell benefits from enhanced individual protection, and redundancy systems can be implemented from one cell to another.

Another possible solution: prefabricated modular data centers (container-type or with assemblable panels). Quick to implement, but less secure, they are particularly suited to temporary uses: relocation of IT equipment during site work, absorption of a sudden capacity increase, activity peak shaving, or rapid go-to-market. They are often favored by telecom operators or certain industrial companies.

However, while modularity is attractive in many respects, it is not the appropriate solution in a number of cases… Thus, unless one has premises or warehouses of sufficient size, it can be difficult to integrate a prefabricated modular data center, along with its associated technical installations, into an existing building. Similarly, a company anticipating an IT capacity increase of 80% in less than two years will not particularly benefit from choosing modular containers, which are more expensive in terms of investment. Finally, security and service level must be taken into account: thus, a company seeking to achieve a Tier 3 or Tier 4 level cannot reasonably opt for a "lightweight" modular solution but must favor concrete and secure compartmentalization, while seeking an eco-efficient and flexible design.

Conversely, when the company does not have strong requirements in terms of service levels (when Tier 2 is sufficient), or when the longevity of the implementation is not sought over a long duration, it is possible to opt for a more economical solution, modular or not.

As we can see, there is no single model. A data center must above all meet the medium- and long-term business constraints and objectives of the company. To make the right choice, the flexibility of solutions as well as investment and operating costs are essential criteria. But overlooking the expected service and security levels, and the specific capacity increase needs of each company, would be a mistake.