Why datacenters are called HKG, SIN and NRT

Open the node list of any international cloud or hosting provider and you will most likely see three capital letters: HKG, SIN, NRT, FRA, LAX. That alphabet was not invented by this industry — it was borrowed, from aviation.

Since the middle of the last century IATA (the International Air Transport Association) has assigned a three-letter code to every airport: HKG is Hong Kong International, SIN is Singapore Changi, NRT is Tokyo Narita. When the datacenter industry needed a short identifier for "a geographic location", it simply adopted a coding system that was already globally current — for practical reasons. It is short (three characters fit any system field), it is unambiguous (one code, one place), and it is naturally friendly to an operations engineer who travels. So "datacenter code = the airport code of its city" became the industry default, and it was kept even where city and airport do not map one-to-one (Tokyo has both NRT Narita and HND Haneda; which one the hosting world uses varies slightly by provider — the Tokyo node in our speed-test matrix is labelled NRT, i.e. the Narita convention).

Cloud vendors brought a second system: Region and AZ. The us-east-1a style is the archetype — a Region is a geographic cluster, an AZ (availability zone) is an independent facility within that region. The letter suffix a/b/c deliberately does not point at a physical location, and that is design, not vagueness: the vendor does not want you binding your workload to "one particular building", but to the abstraction "independence between zones". You request two AZs for redundancy; which two ends of the city they are is the vendor's business. The philosophy is the opposite of the airport code: airport codes want to name a place precisely, AZ naming wants to stay deliberately vague.

The third system is classification: Uptime Institute's Tier I to Tier IV, which sorts datacenters into four grades by redundancy and maintainability. The keyword for Tier III is "concurrently maintainable" (you can service any one power or cooling path without taking the whole facility down); Tier IV is "fault tolerant". The common misuse is treating Tier as a marketing word for "luxury level" — its original meaning is an engineering one with explicit criteria. When you see "Tier IV datacenter" in a sales page, it is worth asking for the certificate number (the common account: certification has a formal process and a public registry).

Three systems, three jobs: airport codes answer "where", Region/AZ answers "how do I survive failure", Tier answers "how reliable". A speed-test tool that names its nodes with airport codes is following that division of labour — a speed-test user cares about "how far is this from my users", which is a purely geographic question, and a geographic question is answered fastest by a geographic code: three letters. Failure domains and grades belong to the selection stage, and the selection pages speak a different language.

Naming is never only a name. Behind the three systems stand three industry viewpoints: aviation gave us geography, cloud vendors gave us abstraction, and the rating bodies gave us engineering standards. Next time you see HKG or us-east-1a, you will know which layer of the world you are looking at.

Write a ~900-word piece of industry lore about the naming systems used for datacenters: why IATA airport codes (three-letter codes like HKG/SIN/NRT) became the industry habit, the logic of cloud vendors' Region/AZ naming (us-east-1a style), and what Uptime Institute Tier I–IV means plus how it is commonly misused. Requirements: for the historical part write only what has public sourcing (mark anything uncertain as "the common account"), and end by explaining, from a product perspective, why a speed-test tool would name its nodes with airport codes.