Exploring the Various Data Center Models

Although businesses continue to migrate more applications and services to the cloud, most also need to maintain a physical data center to ensure the control, reliability, and security necessary for many mission-critical workloads. In fact, ongoing digital transformation efforts are creating strong demand for more data center facilities — analysts expect the global market to double in size through the end of the decade.

Data centers house critical networking, computing, and storage resources, but they come in many different flavors. From massive, town-sized facilities to simple data closets, they provide the space, power, and connectivity necessary to deliver critical data, applications, and services.

Here’s a brief look at the major types of data centers:

Enterprise Data Center

This is what most people think about when discussing data centers — a facility owned and operated by a single company to serve its specific computing needs. These have traditionally been on-premises facilities, but they are often built offsite as well for business continuity purposes. A recent Uptime Institute survey found that privately owned enterprise data centers currently handle about half of all IT workloads.

The major benefit of a privately owned data center is that it gives you full control over your infrastructure and data. That’s particularly important for organizations that rely on proprietary hardware and software specific to their operations. Additionally, direct control over data and infrastructure makes it easier to meet certain regulatory compliance requirements.

Hyperscale Data Center

As the name suggests, hyperscale data centers (HDCs) are enormous facilities used by organizations such as Google, Amazon, and Microsoft with massive data processing and storage requirements. HDCs typically have more than 5,000 servers and occupy 10,000 square feet, although some house hundreds of thousands, even millions, of servers. The world’s largest data center in Langfang, China, occupies 6.3 million square feet, equivalent to 110 football fields.

HDCs aren’t defined solely by size, however. They require the ability to scale compute, memory, networking, and storage resources to meet increasing demand. They can scale either horizontally by adding more servers or vertically by adding more computing resources to existing infrastructure. They must also be able to accommodate increased demands for power, cooling, and real estate.

Colocation Data Center

In a colocation, or colo, arrangement, customers lease space in a provider’s data center for the hardware, software, and system architecture of their choosing. This gives customers the ability to build servers to their specifications, with the CPUs, RAM, and other components of their choice. Colo services include all the necessary power, cooling, and physical security needed, as well as established connections to a variety of telecommunications and network service providers.

There are two models for colocation, retail and wholesale. Generally speaking, retail involves shared space within a data center while wholesale features dedicated space.  

Retail colo is geared towards organizations with relatively modest computing needs. Customers usually have 10 or fewer server racks or cabinets, either dispersed throughout the facility or located in locking cages. Retail colo providers may also offer half racks, quarter racks, or even rack units.

Wholesale facilities are intended for larger enterprises with power requirements of 500kW or more. Wholesale colo customers may lease one or more private, locking cages, a suite, or even an entire facility in a build-to-suit model. Each wholesale customer is kept separate from other customers for greater security.

Conclusion

There’s no one-size-fits-all approach to data centers. Rahi’s data center experts have the expertise to design, build, implement and maintain any type of data center model. We’d welcome the opportunity to discuss your requirements. Contact us to book a 30-minute strategy consultation.