I wanted a Tier One data center.
That sounds backward. Higher tier numbers promised more built-in redundancy and better availability. Why start at the bottom?
Because we wanted the building to be a physical foundation, not an outsourced architecture.
I sent leadership a list of serious providers and asked for any strong preferences or warnings. The requirement was a reliable facility where we could build our own network, power, and systems design on top.
Buying the most elaborate building did not automatically produce the system we needed. It could also hide assumptions inside somebody else's version of redundancy.
I included the tier definitions in the note because the language was easy to misuse. “Tier One” sounded like the best. In this system it was the simplest. That distinction needed to be clear before the vendor conversations started.
The tier chart was clean. Tier One had a single distribution path and basic capacity. Tier Four added fault tolerance, multiple paths, dual power, and very high stated availability.
The actual decision was messier. Location mattered. Carrier access mattered. Staff access mattered. Cost mattered. Most of all, we needed to understand which failures the facility owned and which ones stayed with us.
A data-center tour could show polished floors and impressive generators. I wanted to know where our cables went and what happened when ordinary equipment failed.
I preferred owning the design where our team had the expertise to operate it.
That did not mean ignoring reliability. It meant assembling it deliberately instead of assuming a label had handled the hard parts.
The building would supply space and power. We would bring the rest.
That was the theory. Now we had to choose the building.