When computers were first mass adopted in society, there were mainframes and large consoles were used to access the mainframe. These mainframes were as large as basements in modern homes or even larger; they required (at times) custom, dedicated power lines just to keep them powered.
Furthermore, they were extremely expensive (the Harvard Mark I mainframe….used in the 1940s and later … had a manufacturing cost of $200,000 USD — in 2020, that would be around $3 million USD). These mainframes were used to calculate (think SUPER calculators), primarily using information called data.
These machines were quite big; the Harvard Mark I was 9,500 pounds, or over 4 tons and was over 50 feet long. As more widespread adoption of these units became a reality, these units required massive amounts of customized real estate to house them.
Basically, you needed a large ‘center’ to house these machines that calculated new ‘data.’ Welcome to the idea of a datacenter!
A short, concise understanding of the term datacenter is a large area or room dedicated to housing larger computing devices and the network/electricity/etc. needed to keep them up and running as close to 100% of the time as possible.
Fast-forward to 2020. The typical modern datacenter may have some AS400 units (modern mainframe), but will also have large metal shelfs (called racks) which hold servers, network switches, network routers, network patch panels, backup tape drives, NAS and SAN storage units, and more. The main purpose of all these devices is to do the large calculation, manipulation, and distribution of information for an organization.
Think of it this way:
For most companies, most of the large data sets and information tables stored and updated/calculated against are stored in datacenters. Furthermore, the cloud concept is renting datacenter access from other companies (eg., Microsoft Azure).
To summarize, a data center is the large area or room dedicated to housing larger computing devices and the network/electricity/etc. needed to keep them up and running as close to 100% of the time as possible.