What is hyperscale?
Explore IBM Storage Scale System Subscribe for cloud topic updates
Servers in a hyperscaler

Published: March 12 2024
Contributors: Phill Powell, Ian Smalley

What is hyperscale?

Hyperscale is a distributed computing environment and architecture that is designed to provide extreme scalability to accommodate workloads of massive scale. The related term “hyperscaler” refers to hyperscale data centers, which are significantly larger than traditional on-premises data centers.

As is obvious to anyone working in IT, there are regular jobs and then there are projects whose order of magnitude exceeds normal needs. These supersized cases require extra handling and an expanded sense of proportion. In short, they need the special abilities enabled by hyperscale computing.

Hyperscale computing is the counterpart of using standard enterprise data centers. In hyperscale computing, companies build or help create large and almost infinitely scalable databases.

More to Store with a Global Data Platform

All about IBM’s Global Data Platform—engineered for anyone to access anything anywhere.

Related content

Subscribe to the IBM Newsletter

From data centers to hyperscalers

Hyperscale technologies represent a sea change in how the daily flow and volume of data generated by businesses can be processed.

It may be difficult to remember, but there was a time when a company’s entire data center might have been run from a lone server within an office cabinet.

Then came hypervisors, whose use as abstraction layers let apps in virtual machines (VMs) be relocated from one physical hardware installation to another—a key moment in the rise of hyperscale data centers.

In most cases, the original, on-premises data centers of old simply cannot handle the volume of data now generated, especially that created by hyperscaled applications.

Hyperscale data centers (also called hyperscalers) occupy considerably larger physical space than traditional on-premises data centers, which have tended to be sized somewhere in the 10,000-square-foot range.

According to the IDC definition of a hyperscaler (link resides outside ibm.com), to be considered a true hyperscaler, a company must use 5,000 servers or more and devote at least 10,000 square feet to the operation.

Hyperscale facilities are often multiples beyond that—with building sizes often realizing dimensions approaching 60,000 square feet—roughly the size of a regulation US football field.

And while that may be a typical hyperscale size, it’s not nearly the largest example of storage options. That distinction belongs to the big China Telecom data center (link resides outside ibm.com) located in Horinger, Hohhot, within China’s Inner Mongolia region. This facility, which cost USD 3 billion to construct, covers a staggering 10.7 million square feet and uses 150 megawatts of power. (To visualize a facility of this enormity, try to imagine the combined space of roughly 165 adjoined football fields.)

How do hyperscalers work?

Hyperscalers, an outgrowth of hyperscale computing, are hyperscale data centers primarily used to deliver and manage mega-sized applications.

Hyperscalers are largely cut from the same cloth as traditional data centers—but just geared to a much larger scale than typical on-premises data centers. They do this by building and running an enormous hardware and software infrastructure in the hyperscaler facilities. Millions of servers distributed across many data centers provide seemingly endless storage and computing resources.

Since data traffic can fluctuate wildly—especially when running huge applications—hyperscalers accommodate that traffic and stabilize hyperscale operations when increased demand is exhibited within a computing environment. Hyperscalers do this by essentially serving as a form of load balancer, juggling tasks and re-directing computing resources as needed.

Benefits of hyperscale

Some companies aren’t really poised to take advantage of hyperscale technology, due to its cost of entrance and other associated costs. However, for most businesses, the positive aspects far outweigh those costs.

Managed data centers

Virtualization enables cloud computing, and CSPs host hyperscale data centers to accommodate the many uses of cloud computing and the data it generates. Meanwhile, the user is freed up from the many details and occasional operational headaches of running a data center onsite, and instead interacts with needed cloud resources through application programming interfaces (APIs).

Engineered for big jobs

Hyperscale infrastructure, which underpins hyperscale computing, fosters both high performance and redundancy, making hyperscale projects the perfect choice for cloud computing activities and big data processing.

Efficiency

Because it’s built expressly to efficiently handle hyperscale purposes, hyperscale infrastructure is able to increase the cost-effectiveness of operation, even when handling mammoth workloads.

Enhanced networking

It does little good to build what is essentially a server farm unless the servers there have strong, lightning-quick connectivity (with low latency), so those servers can communicate effectively with each other. Hyperscalers provide that.

Leading hyperscale providers

The hyperscale market is currently dominated by a “big five” of public cloud providers, with each of these cloud service providers (CSP) possessing their own strengths.

Amazon Web Services (AWS)

As of Q1 2023, AWS achieved a 32% market share (link resides outside ibm.com), making it the largest provider of hyperscale cloud services. Its cloud-themed products focus on aspects like storage options, computing power, automation, databases and data analytics.

Microsoft Azure

Many companies are already consumers of one Microsoft product or another, so those businesses may have an established comfort level with the company and its offerings. That includes Microsoft’s enterprise software, which integrates nicely with Microsoft Azure, its hyperscale product. (Market share: 23%)

Google Cloud Platform (GCP)

Google rose to power and prominence through its mastery of data handling and businesses still seek out GCP (about 10% of the market), especially if those companies are eager to engage in advanced analytics and extend their footprint into artificial intelligence (AI).

IBM Cloud

With long-established technology bona fides, IBM Cloud leverages the company’s expertise in many areas, like AI and dealing with enterprise data centers. IBM Cloud delivers comprehensive services for Infrastructure-as-a-Service (IaaS), Platform-as-a-Service (PaaS) and Software-as-a-Service (SaaS).

Oracle Cloud Infrastructure (OCI)

OCI’s advantages include ease of migrating critical enterprise workloads and the ability to construct cloud-native apps. Another selling point for OCI has been its aggressively low pricing, with Oracle claiming to offer the same basic services as AWS at a fraction of the cost.

How do the “Big Five” differ?

The quick overview summary suggests that AWS excels because of its global access and advanced scalability. GCP is attractive to businesses who need top-level data management and want to make forays into capabilities like machine learning, and Microsoft’s Azure provides smooth product integration and enhanced security.

Beyond the top three competitors, IBM Cloud is sparking considerable interest with its current work with AI, which informs its hyperscale offerings. And Oracle’s been serving OCI up as a platform built to save money, and host cloud-native applications.

As is evident, each of these providers differs significantly in their approach and their particular areas of expertise, but they do share similarities. For example, the top three of these providers now offer cloud-native services supporting zero-trust network access (ZTNA) protocols that are designed to offer an alternative to VPNs, which can experience security vulnerabilities.

The economics of hyperscale

Different companies approach how they negotiate their hyperscale needs with different strategies. 

Not all businesses can afford the substantial financial investments needed to set up elaborate hyperscalers or wish to make such a financial commitment. Remember, we’re not simply talking about construction costs. There are massive equipment purchases to be considered as well.

It’s also been observed that some hyperscale facilities use more power than small cities. So, electricity costs and environmental concerns usually must also be factored into the company’s strategy about hyperscaler use.

As an alternative to purchasing equipment, some companies still engage in the practice of colocation, in which servers or other computing equipment are instead rented.

Hyperscalers have also made an impact when it comes to Internet of Things (IoT) devices and how they’re managed. The fact that hyperscale clouds are being used in conjunction with pre-existing equipment (in many instances) helps drive down the price tag of infrastructure investments for the IoT ecosystem—making it a better value for companies.

Related solutions
IBM Storage Scale System

Discover the easiest way to deploy IBM Storage Scale with performance and capacity that scales from 1 to over 1000 nodes.

Explore IBM Storage Scale

IBM storage area network (SAN) solutions

Update your storage infrastructure with IBM b-type and c-type SAN switches and reduce your total cost of ownership.

Explore IBM storage area network (SAN) solutions

IBM Consulting Cloud Accelerator

Shift your adoption of the IBM Cloud platform into overdrive with an Accelerator that supports a wide range of apps and landing zones.

Explore IBM Consulting Cloud Accelerator
IBM Storage

Balance both sides of the data storage equation by using IBM software (like IBM FlashSystem) and IBM hardware solutions.

Explore IBM Storage

Resources What is a storage area network (SAN)?

A storage area network (SAN) is a dedicated network that is tailored to a specific environment—combining servers, storage systems, networking switches, software and services.

What is a data center?

A data center is a physical room, building or facility that houses IT infrastructure for building, running, and delivering applications and services, and for storing and managing the data associated with those applications and services.

What is object storage?

Object storage is a data storage architecture ideal for storing, archiving, backing up and managing high volumes of static unstructured data—reliably, efficiently and affordably.

What is data storage?

Data storage refers to magnetic, optical, or mechanical media that records and preserves digital information for ongoing or future operations.

What is IT infrastructure?

IT infrastructure refers to the combined components needed for the operation and management of enterprise IT services and IT environments.

What is latency?

Latency is a measurement of delay in a system. Network latency is the amount of time it takes for data to travel from one point to another across a network.

Take the next step

Hyperscale has added new dimensionality to storage issues, unlocking entire new realms of opportunity. Find out how to best leverage this powerful new capability in your business with IBM Storage Scale.

Explore IBM Storage Scale