Data centers serve the primary function of processing, storing, and distributing data. Many types of data centers exist to fulfill these specific purposes, from enormous hyperscale facilities built in a modular way to small edge sites, and from private facilities for a single enterprise to the multi-tenancy of colocation.
The five main types of data centers are enterprise data centers, colocation data centers, hyperscale data centers, edge data centers, and modular data centers. Each of these types of data centers vary in terms of their typical size, power capacity requirements, location in the network, and purpose.
Dgtl Infra provides an in-depth overview of the five primary types of data centers, as well as examples of how they are used, who they are operated by, and who they are used by.
1. Enterprise Data Centers
Enterprise data centers are private facilities that are owned and operated by an individual organization to meet its own IT infrastructure requirements. These data centers are particularly well-suited for organizations that require customized networks or can benefit from economies of scale due to significant amounts of traffic or data being processed and managed within their data center.
Organizations may choose to maintain enterprise data centers for a variety of reasons, such as regulatory compliance, privacy protection, superior performance, enhanced security, or cost-effectiveness. These data centers are specifically tailored to be compatible with an organization’s unique applications and processes and may be located either on-premises or off-premises, depending on factors such as power availability, water accessibility, connectivity, and security.
In an enterprise data center, an in-house IT department manages the servers, storage, and networking equipment. Meanwhile, the power equipment, cooling infrastructure, and security systems may be managed either internally or outsourced to a third-party through a facility management contract.
Organizations are increasingly adopting a hybrid IT model, combining cloud services with data center operations both on-premises and via colocation data centers. Enterprises are transitioning from on-premises to cloud environments primarily for enhanced scalability and to expedite the launch of new applications and services. Moreover, enterprise data centers often face issues with resource underutilization, stemming from inadequate resource allocation, lack of workload optimization, and limited visibility into resource usage.
Examples of Enterprise Data Centers
Enterprise data centers are used by large multinational corporations, examples of which include Bank of America, Coca-Cola, ExxonMobil, JPMorgan Chase, Procter & Gamble, and Walmart. Examples of why an organization would use an enterprise data center include:
- Privacy and Regulatory Compliance: Enterprise data centers are often used to store and manage data generated by an organization’s operations, such as customer information and sensitive data like financial records. These facilities enable organizations to adhere to regulatory compliance requirements, including PCI-DSS, HIPAA/HITECH, FedRAMP, and GDPR, amongst others
- Disaster Recovery: Enterprise data centers can provide redundancy and high availability, making them ideal for supporting business continuity in the event of a natural disaster or power outage
2. Colocation Data Centers
Colocation data centers, also known as multi-tenant data centers (MTDCs), are used by multiple organizations – often 20 or more – to house their computing hardware, servers, and supporting infrastructure, such as power, cooling, and networking equipment, in an off-site location. These third-party facilities are particularly useful for organizations that lack the space or IT resources to manage their own enterprise data center, freeing up IT personnel and financial resources for other initiatives.
Colocation data centers allow organizations to lease the space they require to host their data, with the ability to quickly scale up or down to meet changing needs. These facilities provide organizations with high availability (uptime), robust bandwidth capacity, and low latency access to data. To meet these expectations, colocation data centers regularly refresh and upgrade their hardware and technology.
Providers of colocation data centers, such as Equinix and Digital Realty, offer organizations greater operational flexibility, diverse connectivity options, and low latency access to cloud on-ramps, while reducing an organization’s capital requirements.
Colocation facilities vary in size and cater to different data center infrastructure needs, with retail colocation data centers serving smaller organizations and wholesale data centers accommodating larger enterprises. Additionally, managed services data centers are facilities that provide organizations with various IT services, including server management, storage management, and network management, all delivered by a third-party provider.
Below is further detail on each of these types of colocation data centers:
Retail Colocation Data Centers
Retail colocation facilities provide a turn-key data center service that caters to customers with smaller power capacity needs within the same data halls. The data center operator manages day-to-day operations, power capacity, cooling, and security, as well as access to telecommunications carriers and internet service providers (ISPs). Customers are only responsible for maintaining their equipment within cabinets or cages.
Retail colocation services support smaller data center footprints and lower individual power capacity requirements, usually from 100 kilowatts (kW) to 1 megawatt (MW). These contracts typically last between 1 to 3 years and take the form of leases or license agreements.
Examples of retail colocation data center providers are Equinix, CoreSite, and Cyxtera.
Wholesale Data Centers
Wholesale data centers are facilities leased to a single customer, who is responsible for building out and managing the day-to-day operations of either a full building or data hall to create a fully functioning data center. Unlike retail colocation providers, wholesale data center operators offer only space and power without additional services. Also, wholesale data centers have fewer connectivity options to carriers and internet service providers (ISPs).
Wholesale data centers are designed to accommodate large enterprises, government agencies, and retail colocation providers with white space rooms ranging from 10,000 to 20,000 square feet or larger. These facilities are leased on longer terms, usually ranging from 5 to 10 years, and are suitable for individual power capacity requirements of 1 to 5 megawatts (MW).
Examples of wholesale data center providers are Digital Realty, CyrusOne, and NTT Global Data Centers.
Managed Services Data Centers
Managed services data centers, typically offered by managed service providers (MSPs), involve third-party management of data center infrastructure, including servers, storage, networking, and security systems, offering continuous monitoring and maintenance by skilled professionals. These services allow organizations to outsource IT operations, comprising server and storage management, facilities operations, with adaptability across various setups like on-site, colocation, or cloud data center environments.
The goal of managed services data centers is to reduce or eliminate an organization’s IT staff workload. They enable organizations to focus on strategic objectives and application management, while ensuring data center performance and reliability through service level agreements (SLAs).
Examples of managed services data center providers are Flexential and TierPoint.
3. Hyperscale Data Centers
Hyperscale data centers, also known as cloud data centers, are massive, centralized, and custom-built facilities that are operated by a single company. These facilities support primarily cloud service providers (CSPs) and large internet companies with enormous compute, storage, and networking requirements. In terms of footprint, hyperscale data centers house thousands of racks and tens of thousands of servers across 50,000 to 1+ million square feet. As such, these data center facilities can cost upwards of $1 billion to construct.
Hyperscale data centers are located on the outskirts of major cities, providing supply and cost advantages in land and power. These data storage facilities are leased on very long terms, usually ranging from 10 to 15+ years. They are designed to facilitate scalable power capacity, accommodating requirements from 5 to 100 megawatts (MW).
Geographically, hyperscale data centers are built in “Tier 1” data center markets. In the United States, these markets include Northern Virginia, Dallas, Northern California (Silicon Valley), Phoenix, Chicago, Atlanta, and Portland / Hillsboro, amongst others. While in Europe, the Tier 1 data center markets are Frankfurt, London, Amsterdam, Paris, and Dublin. In Asia Pacific (excluding China) the top data center markets include Tokyo, Singapore, Sydney, and Hong Kong.
Hyperscale data centers, given their size and locations, also incorporate many efficient operating strategies including server virtualization, software-defined networking (SDN), and much lower power usage effectiveness (PUE) ratios, as compared to other types of data center infrastructure.
Examples of Hyperscale Data Centers
Hyperscale data centers can be insourced or self-built by the “hyperscalers”, which include companies such as Amazon Web Services (AWS), Microsoft Azure, Google Cloud, and Meta Platforms (Facebook). Alternatively, these companies can outsource and lease capacity for their cloud data centers from providers such as Digital Realty, CloudHQ, and STACK Infrastructure. For cloud service providers (CSPs) like AWS, hyperscale data centers form the foundation of their cloud regions and availability zones.
4. Edge Data Centers
Edge data centers, also known as micro data centers, are smaller, decentralized facilities that provide compute and storage in a location closer to where data is being generated and used. They are situated near their intended users, allowing for real-time data processing and analysis. Because edge data centers perform this analysis closer to the source of data creation, response latency is reduced and bandwidth is optimized, which facilitates the development of new applications. This is in contrast to traditional centralized data storage facilities that rely on multiple networks and data centers for processing.
The term “edge” refers to the location at which these types of data centers are typically deployed, which is closer to the point of connectivity. Edge data centers can be deployed as standalone facilities or in a variety of environments, such as at telecommunications central offices, cable headends (i.e., local distribution points), the base of cell towers, or on-premises at an enterprise.
In terms of scale, edge data centers serve individual power capacity requirements from 50 kilowatts (kW) to 500 kW of power capacity, or 5 to 50 racks. Multiple edge data centers can be interconnected to increase capacity, improve resiliency, and enable workload migration within a local area.
The growth of edge data centers is being driven by the demand for 5G connectivity, the proliferation of the Internet of Things (IoT), artificial intelligence (AI), big data analytics, and automation. By processing data closer to its source, edge data centers can support these emerging technologies.
Examples of Edge Data Centers
Examples of edge data center providers that have metro edge and mobile (or micro) edge facilities include:
- Metro Edge: AtlasEdge, Cologix, DataBank, EdgeConneX
- Mobile Edge (Micro Edge): EdgePresence, Vapor IO
Metro edge providers are larger, supporting individual power capacity requirements of 100 kilowatts (kW) to 500 kW or approximately 50 racks, while mobile (or micro) edge providers are smaller, suitable for individual power capacity requirements of <100 kW or approximately 5 racks.
READ MORE: What is an Edge Data Center? (With Examples)
5. Modular Data Centers
Modular data centers are standardized buildings that are pre-engineered and prefabricated, complete with power and cooling infrastructure, and used to house computer servers and network equipment. They come in various types, including container data centers, prefabricated data halls, and prefabricated power & cooling modules.
The two primary purposes of modular data centers are to reduce the time and cost required for constructing and deploying a data center. Modular data storage facilities save time by shifting conventional on-site construction tasks to off-site manufacturing facilities, while they deliver cost savings by using standardization and reducing on-site labor.
Below are further details on the different types of modular data centers:
Container Data Centers
Container data centers, also known as containerized data centers, are “all-in-one” standardized units that combine the facility structure, IT equipment, power infrastructure, and cooling systems into one cohesive enclosure. For example, these data center components include the power distribution units (PDUs), uninterruptible power supply (UPS) with batteries, and air- or water-cooled chillers.
A container data center can accommodate various designs, with IT capacity ranging from 20 kilowatts (kW) to 250 kW and 4 to 20 racks.
Prefabricated Data Halls
Prefabricated data halls are a modular data center deployment type that enables the rapid addition of ‘white space’ to an existing data center facility by using pre-manufactured components. These data halls, consisting of several thousand square feet of space and 50 to 200+ racks, facilitate the construction of large wholesale and hyperscale data centers through the assembly of smaller subsystems.
Prefabricated Power and Cooling Modules
Prefabricated power and cooling modules are comprised of fully pre-engineered, integrated, and pretested power or cooling systems, which are manufactured off-site. Once completed, these modules are shipped to a data center and are immediately ready for installation.
An enclosed power module has a capacity range of 250 kilowatts (kW) to 500 kW, whereas a skid-mounted design can support higher capacities of 1 to 2 megawatts.