Data Centers Explained: Types, Features, and Choosing the Right Model

Whether you’re a new business setting up its first data center (DC) or an existing organization needing to expand its DC locations, it’s valuable to understand exactly what a data center is and what your options are for DC deployment when you need to add or move a data center.

This article presents a clear idea of what a data center is, the different types of data center models, and what’s needed to create a data center.

(Related reading: data center security & data center energy optimization.)

What are data centers?

The term “data center” refers to the physical facility where large-scale computing systems and infrastructure is stored and deployed.

Key components of the data center include:

• Server and data storage systems.
• Network communication devices.
• HVAC.
• Power systems.
• The building structure itself, ideally designed for optimal, secure computing performance and energy consumption.

Data centers house and run the critical IT infrastructure that large and small organizations use to run their businesses, process data, and safely connect to the outside world.

Who uses and runs on data centers?

Data centers are used by a wide variety of organizations and industries, including government agencies, financial institutions, healthcare providers, educational institutions, retail companies, and technology enterprises.

These facilities are essential for:

• Hosting websites.
• Managing cloud applications.
• Supporting e-commerce platforms.
• Processing large-scale data.
• Enabling communication services.

The data center may be owned and operated by an organization for its internal IT workloads or operated by a data center services provider. The vendor may offer the data center space to store and deploy computing systems as a managed service provider or a colocation service or operate a cloud-based environment to deliver computing services on a subscription basis to end-users over the web.

Where are data centers located?

Data centers are strategically located based on several factors, such as proximity to end-users, access to reliable power and cooling, connectivity to robust fiber-optic networks, and the availability of land and resources.

Large data centers are often found in regions with cooler climates to reduce cooling costs and areas with low risk of natural disasters. Examples include rural areas, suburban locations, and even urban hubs, depending on the needs of the organization.

(Related reading: disaster recovery plans.)

Additionally, edge data centers are positioned closer to end-users to support low-latency applications, while hyperscale data centers are distributed globally for maximum scalability and redundancy.

Types of data centers

Data centers are housed in secured environments with controlled access. The six common data center types are:

1. Enterprise data centers.
2. Managed service(s) data centers.
3. Colocation data centers.
4. Cloud data centers.
5. Hyperscale data centers.
6. Edge data centers.

1. Enterprise data center

An enterprise data center is typically owned by a single organization or enterprise with predictable usage and limited scalability requirements. These data centers may be located on-premise or off-site and can also operate as private cloud data centers.

Managed internally, enterprise data centers are generally used to handle the organization’s internal data and IT workloads. They are best suited for organizations with sufficient resources and expertise to build, operate, and maintain their own facilities.

2. Managed services data center

A managed service data center is usually located off-site and operated by a Managed Service Provider (MSP). While the maintenance and operational responsibilities are outsourced to the MSP, a single organization may still own the rights to use the facility.

These data centers are designed for predictable usage, making them less scalable than other models. Managed services data centers are ideal for organizations with large-scale computing needs but lacking the internal expertise and resources to independently operate and maintain a facility.

3. Colocation data center

In a collocated data center, several organizations share the physical space while managing their own computing infrastructure. The building infrastructure, including cooling, HVAC, and power, is operated by a third-party MSP or vendor. Colocation data centers offer higher scalability, attracting a diverse customer base with varying usage demands.

This model is particularly suitable for organizations that have the expertise to manage limited data center operations but lack the capital expenditures (CapEx) and resources to build and operate an entire facility.

(Related reading: data center colocation.)

4. Cloud data center

A cloud data center is a collection of distributed facilities designed to provide global cloud-based services. These services are offered on-demand and on a subscription basis, commonly as:

• Software as a Service (SaaS).
• Infrastructure as a Service (IaaS).
• Platform as a Service (PaaS).

(A complete overview of SaaS, PaaS, & IaaS.)

Highly scalable and fully managed by the vendor, cloud data centers enable organizations of all sizes to pay for the computing resources they need, when they need them, and at the scale required for their operations. This model is suitable for businesses looking for cost-efficient, flexible, and scalable data center solutions.

5. Hyperscale data centers

Hyperscale data centers are the largest players in the industry, operated by technology giants like AWS, Microsoft Azure, and Google Cloud. These facilities offer a variety of cloud services with unparalleled scalability.

Globally distributed and fully automated, hyperscale data centers are managed by large technology enterprises that have the resources, expertise, and CapEx to build and maintain extensive networks of data centers worldwide. This model is ideal for tech companies aiming to compete in the cloud computing space.

6. Edge data center

Lastly, edge data centers are located close to end-users, such as individual consumers of telecommunications services. These facilities are often managed and operated by ISPs and telcos.

Edge data centers are used for applications requiring real-time processing, IoT, and low-latency workloads. This model is particularly well-suited for organizations providing Internet-based services where user experience and networking performance are critical to business success.

Standards for data center performance

When evaluating data center options, DCs can be ranked by tiers for their potential infrastructure performance (uptime). The Uptime Institute classifies data centers according to the following tiers:

• Tier I Basic capacity: Supports IT technology for an office setting.
• Tier II Redundant components: Provides redundancy for power and cooling for better maintenance and protection against disruptions.
• Tier III Concurrently maintainable with redundant components and distribution paths: Building on Tier II, these facilities require no shutdowns when equipment needs maintenance or support.
• Tier IV Adds fault tolerance to Tier III. All systems require a fault-tolerant design and continuous cooling to make the environment stable.

Data center power usage effectiveness (PUE)

An important consideration for the data center design is its Power Usage Effectiveness (PUE). It is defined mathematically as: Total Facility Energy/IT equipment Energy.

It is a measure of data center energy consumption that describes the ratio of how much energy is consumed by the data center to fulfill all energy requirements of the computing systems operating in the data center. Ideally this ratio is 1.0, but typically averages 1.58 in recent years. The additional energy demand comes primarily from HVAC, building energy and security systems.

Support infrastructure is crucial for ensuring data center functionality and includes:

• Secured locations protected against unauthorized access.
• Visitor logging to meet auditing requirements. (Learn more about audit logs.)
• Redundant power supplies.
• Fire suppression systems.
• Climate control systems for equipment protection.

Latest trends

Recent trends around renewable energy and high demand for scalable computing workloads such as HPC and AI systems is driving advancement in data center technologies for improved PUE.

AI methods such as reinforcement learning – similar to ones used to develop ChatGPT – are being used to optimize energy consumption for data centers. This has resulted in the following trends:

• Lower greenhouse emissions.
• Improved energy consumption by 20%.
• Reduced water usage by 4%.

At a time when demand for data center resources is growing around 20% annually, the energy consumption is expected to reach around 300 gigawatts by the year 2030.

Choosing your data center type

Selecting the right data center type depends on your organizational needs and constraints. Your organization may have built an enterprise DC decades ago, but now it wants to move to a cloud service provider (CSP) as it rolls out new applications across the world.

Or after an acquisition, you may want to convert a CoLo DC to an MSP or consolidate an acquired CoLo into your existing enterprise data center. As noted earlier, many organizations have multiple data centers at various locations. You can mix and match data center types, choosing the DC model that fits the needs of each specific situation.

There is nothing constraining you from using different DC types for your unique needs. The key is knowing what your data center options are so you can choose the right DC model as your needs change.