There’s not a shortage of terms to go around when it comes to talking about prefabricated data centers. Terms like containerized, modular, self-contained, portable, pods, skids, and mobile, to name a few. But all of these terms can create quite a bit of confusion when trying to understand if prefabricated is right for you, and if it is, which “flavor”.
We just released a white paper to help sort this out. White Paper 165, titled “Types of Prefabricated Modular Data Centers” proposes standard terminology (simple, descriptive terms) for categorizing the types of prefabricated modular data centers that exist, and defines and compares their key attributes. The paper also provides some guidelines to help choose the best approach(es) based on your business requirements. In a nut shell, the paper proposes defining prefabricated data center systems (1) by function, (2), by form factor, and (3) by configuration.
There are three core functions (called functional blocks in the paper): power, cooling, and IT. As you might imagine, power includes systems like switchgear, ATS, UPS, batteries, transformers, and panelboards. The cooling components depend on the architecture selected, whether it be a chiller design, DX cooling plant, or indirect or direct fresh air system. And the IT function includes all of the systems generally placed in the “white space” to support the IT equipment, like the racks, power distribution, air distribution (CRAHs, ducts), lights, security, and fire protection.
As for form factor, the majority of prefabricated data centers use ISO containers, enclosures, or are skid-mounted. The form factor influences the cost of the system, the placement of the system, and the flexibility in terms of capacity and layout within the module. These and other pro’s and con’s are discussed in greater detail within the white paper.
Lastly, we’ve identified configurations, or approaches to implementing prefabricated modules. A data center can deploy a completely prefabricated data center – including the power plant, the cooling plant, and the IT space – we call this configuration a “fully prefabricated data center”. This makes sense when scalability is a key business driver, for multi-tenant data centers that want to partition their IT spaces by tenants, for large scale homogeneous IT data centers that want to step & repeat as the load grows, and for high density, HPC computing data centers that want to scale as their IT server need scales.
But there are other situations where using some prefabricated systems and some traditional systems make sense. In the paper, this is called a “semi-prefabricated data center”. For example, let’s say an existing data center has plenty of IT space to continue to grow their IT load, but they are constrained on power and/or cooling capacity. Deploying prefabricated power and cooling can extend the life of the data center. The last configuration, called “all-in-one data centers” is really a subset of the fully prefabricated data center, but we’ve called it out as a separate configuration due to its unique characteristics. In an all-in-one, multiple functions are co-located in a single structure. All-in-ones are well-suited for temporary needs like military use or sporting events, for data centers that require mobility, or for small branch sites or disaster recovery sites.
The white paper is filled with images that illustrate what the different functional blocks and form factors look like, as well as layouts of the common configurations. Ultimately, the goal is to be able to converse with some standard lingo so it is explicitly clear what we’re talking about when we talk about prefabricated data centers.