In the world of data centers, "hyperscale" belongs to the same category as words like "high performance" and "high availability": Folks toss it around frequently, but rarely define exactly what it means.
What is a hyperscale data center? What differentiates the hundreds of hyperscale facilities in existence from the opposite of hyperscale facilities? What is the opposite of a hyperscale data center, anyway?
I can't offer definitive answers to those questions because, as I'll explain, there is no universal definition for a hyperscale data center. However, in this article, I can provide some guidance on how to think about the defining characteristics of hyperscale data center facilities.
Defining hyperscale
To understand what a hyperscale data center is, let's start by exploring the meaning of hyperscale.
There's no "official" definition of the term. But if you poke around the internet, you'll notice that popular interpretations of hyperscale commonly emphasize a few core characteristics:
- Scalability: Unsurprisingly, most definitions of hyperscale underline the importance of being able to scale infrastructure easily. For example, Digital Realty says that being hyperscale means you can deploy "extra resources without requiring additional cooling, electrical power, or physical space."
- Distributed computing: The idea of distributed computing architectures — meaning ones where workloads are shared across multiple servers — is frequently associated with hyperscaling. HPE, for instance, says that a hyperscale architecture involves the use of a distributed computing environment.
- Importance: According to companies like Parallels and Vertiv, hyperscaling is associated with "mission-critical" or "business-critical" workloads. From this perspective, workloads that aren’t as important to the business don't qualify as hyperscale workloads.
These concepts get us closer to an understanding of what hyperscale means. Unfortunately, though, there are seemingly no definitions of the hyperscale concept that offer any hard numbers. This lack of a clear definition raises other critical questions. How quickly do you need to be able to add resources to your infrastructure in order to say you are hyperscaling, as opposed to doing "normal" scaling? Which quantity of CPU, memory, or other resources do you need to be able to scale in order to qualify as a hyperscaler rather than a "regular" scaler? How much revenue needs to be tied to a workload in order for it to count as a mission-critical, hyperscale workload (if being mission-critical is even an intrinsic part of the meaning of hyperscale)?
Instead of answering questions like these, most takes on the meaning of “hyperscale” employ generic comparative terms, such as "high levels of performance" and "high availability" (to quote again from HPE's definition).
A market-based definition of hyperscale data centers
Another approach to defining hyperscale data centers is to think not in terms of how they operate, but who owns them.
In most cases, hyperscale data centers are associated with very large, tech-centric companies, such as Google and Amazon. Smaller organizations may also deploy highly scalable infrastructure, but they rarely feature in conversations about hyperscalers. Neither do larger companies (like big banks, for example, which also operate their own data centers in many cases) that might have massively scalable data center infrastructures, but which aren't primarily in the technology business.
Perhaps we should think of hyperscale data centers, then, as being defined in terms of how much revenue a company generates through its data centers, or even how many employees it has. Those characteristics have nothing to do with what's actually in a data center or how rapidly it can scale from an operational standpoint. But again, no one's precisely defining hyperscale data centers based on the latter characteristics, either.
Hyperscale's meaning is in the eye of the beholder
If there's any conclusion to be drawn about the definition of hyperscaling or hyperscale data centers, it's that there's no clear definition. The data centers that we think of as hyperscale facilities are perhaps distinguished more by the companies that own them — which are, again, large tech companies in most cases — than by any intrinsic technical characteristics.
So, if you want to be a hyperscaler, don't go out and obsess over how to build server architectures and infrastructure orchestration tools that allow you to scale massively. Focus instead on becoming a large technology company, at which point you'll arguably be considered a hyperscaler regardless of whether you actually do anything different from other organizations that also operate massively scalable infrastructures.
