More than a decade ago, we started our How to Speak Like a Data Center Geek series. The idea was to help more people understand the terminology used by those of us who think about data centers every day.
Over the years, How to Speak Like a Data Center Geek has been among the most popular series on this blog. Even today, many people are still reading posts from the very early days of the series, written way back in 2013. We’re glad that people are still interested in those older blogs, but the industry has changed so much since they were published, and we don’t want to give readers an outdated view of things.
With that in mind, we’re taking a cue from the Hollywood movie studios by rebooting How to Speak Like a Data Center Geek. We’ll take another look at some of the terms from those original posts and define some new terms that have popped up in the years since. Let’s start with the basics: defining the different types of data centers and the common structural elements.
Colocation (“colo” for short): When an organization deploys its hardware in a data center owned and operated by a third party. Customers lease space, power and cooling capabilities from the colocation provider; therefore, they’re commonly called “tenants,” just like in any other real estate lease. Since colocation data centers are shared between multiple customers, they’re also referred to as “multitenant” data centers.
On-premises (“on-prem” for short): Data centers owned and operated by an organization for their own private use. Over the years, it’s become untenable for enterprises to continue using on-premises data centers exclusively. Leading colocation providers constantly invest to update their facilities, thus ensuring their customers always have access to the latest data center capabilities. On the other hand, updating an on-premises data center to enable next-generation technologies is a headache that many business leaders would just as soon avoid.
Learn more about on-premises versus colocation.
Rack: An open metal frame for storing and organizing data center equipment. Racks are designed for mounting multiple units of IT hardware.
Cabinet: A closed metal enclosure for storing data center equipment. Like racks, cabinets hold multiple servers and other IT devices, but unlike racks, cabinets are closed on all sides, with doors, side panels, and a top cover. Cabinets can have integrated features for power distribution, cooling, networking and cable management.
Learn more about secure cabinets.
Cage: A secure storage area that subdivides a colocation space and holds multiple racks and/or cabinets. Cages can be private or shared. They offer an extra layer of security for customers, have mesh walls to help with cooling efficiency, and are interconnected to other parts of the data center.
High-performance data center: A colocation data center designed to help customers tackle their biggest priorities, including connecting with partners and service providers easily, adopting AI, implementing a hybrid multicloud architecture and meeting their sustainability goals. The opposite of commodity colocation data centers that provide the basics (low-cost space and power) but nothing else of value.
Learn more about high-performance data centers.
IBX: The brand name Equinix uses for its colocation data centers. It’s short for International Business Exchange™, which speaks to the unique value our data centers provide. Ever since Equinix was founded more than 25 years ago, our data centers have been digital gathering places. Thousands of enterprises and service providers come together here to interconnect and exchange data with one another. This empowers digital ecosystems that drive business value for all participants.
Today, there are 260+ Equinix IBX® data centers in 70+ markets worldwide. This is up from only 95 data centers when we first defined the term back in 2013.
Hyperscale data center: A specific type of colocation data center that’s built to meet the needs of very large customers, such as major clouds and other global service providers. Everything about hyperscale data centers is big, from the square footage to the datasets processed there. They play a key role in enabling AI adoption, as they provide the massive space and power capacity required to train large language models (LLMs).
Although enterprise customers typically won’t deploy in hyperscale data centers themselves, they will benefit from them indirectly. For instance, to jump-start their AI strategies, they may use publicly available LLMs that were trained in hyperscale data centers.
Learn more about hyperscale, and how it differs from conventional colocation.
xScale: The Equinix portfolio of hyperscale data centers. Equinix xScale® data centers are available in strategic global metros such as London, Sydney and São Paulo. In 2024, we announced a joint venture intended to raise more than $15 billion in capital to fund new xScale data centers in the United States. We’re doing this to help meet growing demand for hyperscale services in more locations.
That’s it for this installation of How to Speak Like a Data Center Geek! Going forward, we’ll continue to bring you more terms used in the industry and share what they mean for your business. You’ll be able to find those new posts—along with some older ones—on the series landing page.
For now, learn how to migrate to a high-performance data center.
Subscribe to the Equinix Blog