Data Centre Colocation: A Practical Guide for Growing Businesses

What is colocation? Colocation, often shortened to "colo", means placing servers and network equipment that you own in a third-party data centre. A data centre is a purpose-built facility for IT infrastructure. In a colocation arrangement, your company continues to own and manage the hardware, while the provider operates the building and supplies power, cooling, physical security and connectivity.

This model is the middle ground between on-premises infrastructure, where your organisation owns both the equipment and the facility around it, and public cloud, where computing resources are supplied as a service. This article aims to explain what colocation is, how a facility works, which costs and trade-offs to expect, and which companies benefit most from retaining hardware control without operating their own data centre.

What is a colocation data centre?

A colocation data centre is a third-party facility where organisations rent space for their own servers, storage systems and network equipment. The provider supplies and operates the supporting infrastructure, including the building, electrical power, cooling, physical protection and network connectivity. Your company purchases, configures and maintains the IT hardware and decides which operating systems, applications and security controls run on it.

Space is commonly measured in rack units or rented as part of a cabinet, a full cabinet or a larger private area. A rack is a standardised frame or enclosure used to organise servers and other equipment vertically. Because customers share the facility infrastructure while retaining control over their own hardware, colocation can provide access to capabilities that would be expensive to reproduce in an office server room or private facility.

The boundary of responsibility matters, as colocation does not usually mean that the provider manages the customer’s operating system or applications. Managed services can be added under a separate agreement, but the basic colocation service concerns the facility and the physical environment around the equipment.

How does a colocation data centre work?

You select the required space and power allocation, install the equipment in the facility and connect it to the chosen networks. The provider then keeps that equipment powered, cooled, physically protected and connected according to the service agreement.

Facilities use redundant systems to reduce the effect of a component failure or planned maintenance. N+1 redundancy means the facility has the capacity required for normal operation, represented by N, plus one additional component. An uninterruptible power supply, or UPS, supplies short-term power when the main electrical source is interrupted, while backup generators support longer outages. Facility resilience is often summarised in data centre tier levels, from Tier I to Tier IV, which are covered in the benefits section below.

What the provider supplies vs what you own

Remote hands is an on-site support service through which data centre technicians perform agreed physical tasks, such as checking an indicator, replacing a cable or restarting equipment, when your team cannot visit the site.

Inside the facility: power, cooling, security, and connectivity

• Power: Utility feeds, UPS systems, power distribution equipment and backup generators help maintain supply during interruptions and maintenance.

•  

• Cooling: HVAC systems control temperature, airflow and humidity so that equipment operates within safe environmental limits. Redundant cooling capacity can keep the IT environment stable if one component is unavailable.

•  

• Physical security: Measures can include staffed access points, 24/7 video surveillance, access logs, electronic locks and optional biometric access. The exact controls vary by facility and service level.

•  

• Connectivity: You can connect to network operators and partners through a cross-connect, which is a dedicated physical cable link inside the facility. A carrier-neutral data centre allows you to choose among multiple telecommunications carriers instead of requiring a single network provider.

Types of colocation: retail, wholesale, and how space is rented

Colocation contracts generally fall into retail and wholesale models. The practical difference concerns the amount of space and power committed, the way services are bundled and the level of customisation available.

Retail vs wholesale colocation

Retail colocation serves smaller deployments through individual rack units, partial cabinets, full cabinets or a limited number of racks. Wholesale colocation is designed for larger requirements, often with dedicated power capacity and a private cage, suite or data hall.

Racks, cabinets, cages, and suites

• Rack unit (U): The smallest standard measurement for rack-mounted equipment. One rack unit equals 1.75 inches, or 44.45 mm, of vertical space.

•  

• Rack or cabinet: A frame or enclosed unit that houses multiple servers, storage systems and network devices. A common full-size cabinet provides 42U of usable vertical space.

•  

• Cage: A physically separated and access-controlled area containing several cabinets within a shared data hall.

•  

• Suite: A private room or larger dedicated area that can accommodate multiple rows of racks and customer-specific security or infrastructure requirements.

Benefits of colocation

Colocation gives an organisation control over its hardware without requiring it to build and operate the facility that supports that hardware. The value comes from sharing specialised infrastructure while keeping ownership of the servers and their configuration.

• A different cost structure: You avoid the capital expense of constructing a data centre and move facility costs into a more predictable operating expense. Hardware purchase and replacement costs still remain with your company.

•  

• Infrastructure resilience: Redundant power, cooling and connectivity reduce single points of failure. The Uptime Institute classifies data centres into four tiers according to their infrastructure resilience. Tier I provides basic capacity, Tier II adds redundant power and cooling components, Tier III allows planned maintenance without interrupting IT operations, and Tier IV adds fault tolerance, so that a single equipment failure or distribution path interruption does not affect operations. Tier III and Tier IV are commonly associated with 99.982% and 99.995% availability, respectively. These percentages are reference values rather than guaranteed outcomes, as tier classification evaluates infrastructure design and operational capability.

•  

• Physical protection: Purpose-built facilities can provide continuous surveillance, controlled access, fire detection and environmental monitoring that would be costly to reproduce in a normal office.

• Connectivity choices: A well-connected or carrier-neutral site can provide access to several carriers, private connections and routes designed for consistent network performance.

•  

• Hardware control: Your company chooses the processors, storage, network equipment, operating systems and refresh schedule instead of consuming a standard public-cloud configuration.

•  

• Capacity for growth: Additional rack space, power and connectivity can be contracted as demand grows, subject to availability at the chosen facility.

•  

An SLA, or service-level agreement, sets out measurable commitments such as service availability, response times and support responsibilities. You should evaluate the SLA alongside the facility tier because the two describe different aspects of the service.

Colocation vs cloud, on-premises, and other options

Colocation is the middle option between owning the entire physical environment and renting computing resources as a service.

On-premises infrastructure gives the organisation responsibility for both hardware and facility operations. Public cloud providers such as AWS, Microsoft Azure and Google Cloud supply virtualised resources on demand. Colocation retains customer-owned hardware while transferring facility operations to a specialist provider.

Managed hosting is an option in which the provider supplies and manages the server infrastructure for your company. Edge computing processes data closer to the users or devices that generate it, which can reduce latency. A hybrid cloud combines private infrastructure, including servers hosted in a colocation facility, with public-cloud services. Applications and data can then run in the environment that best matches their requirements for control, performance, cost and scalability. Hybrid infrastructure is already common. Uptime Institute’s 2025 survey found that 45% of respondents’ IT workloads still resided in corporate facilities, with the remainder hosted off-premises.

Which companies benefit most from colocation?

Colocation suits organisations that have a practical reason to own physical infrastructure but do not want to operate the building around it. Company size is less important than workload characteristics. A growing SME may need one cabinet for stable applications, while an enterprise may require several racks, a cage or a private suite.

The strongest candidates usually have existing hardware investments, predictable workloads, performance requirements or governance rules that call for physical control. A small startup with limited and highly variable demand may gain more flexibility from public cloud because it can provision resources without buying servers.

• Financial services: Banks, payment companies and trading organisations may need low, stable and measurable latency. Latency is the time required for data to travel from one point to another. Cross-connects can create direct physical links to network providers, partners or trading venues, while controlled facilities support reliability and regulatory processes.

•  

• IT and telecommunications: Network operators, managed service providers and technology companies use colocation facilities as interconnection points and as locations for servers, storage and network equipment. It is the largest colocation end-use category, at 29.7% of the market (Grand View Research, 2025). The model allows them to reach carriers, partners and customers without constructing a facility in every market.

•  

• Healthcare and other regulated industries: Organisations handling sensitive data may need documented access controls, physical separation and auditable operating procedures. It is the fastest-growing colocation end-use category (Grand View Research, 2025). Requirements can relate to frameworks such as HIPAA, PCI DSS or ISO/IEC 27001, depending on the organisation, jurisdiction and data involved. A dedicated cage or suite can add physical separation, but using a certified facility does not make the customer automatically compliant. The complete technical and organisational setup must satisfy the applicable requirements.

•  

• AI and high-performance computing: GPU systems and high-performance computing can require more power and cooling per rack than standard enterprise equipment. ENCOR Advisors reported in 2026 that high-density racks typically fall in the 10 to 30 kW range, though the appropriate design depends on the hardware and cooling method. Colocation can provide access to higher-density infrastructure when an internal server room cannot safely supply or remove the required power and heat.

•  

• Media, gaming and content delivery: Streaming platforms, game services and content delivery networks depend on bandwidth and responsive connections to users. Distributed infrastructure can place equipment in several regions and create shorter, more predictable network paths.

•  

• Disaster recovery and geographic expansion: A separate colocation site can host recovery infrastructure at a physical distance from the primary environment. It can also give the company a presence in a new region without waiting to design and build its own data centre. These are architecture choices rather than automatic outcomes of colocation, so the organisation still needs a tested recovery plan, appropriate data replication and suitable network routes.

•  

Colocation is particularly useful when the organisation can forecast the workload, needs direct control of the equipment and has people or service partners who can administer the hardware. If demand changes sharply from week to week or the organisation has no reason to own servers, cloud or managed hosting may be a better starting point.

How much does colocation cost?

There is no universal monthly price for colocation, as quotes combine the physical footprint with committed power, connectivity and support, while local electricity prices, facility availability, resilience level and contract term also affect the result. Per-kW pricing means that your company pays according to an agreed amount of electrical capacity, measured in kilowatts, rather than only by rack space.

Typical charges include:

• rack units, a partial cabinet, a full cabinet, a cage or a suite;
• committed power capacity and actual power use, depending on the contract;
• bandwidth, IP resources and network services;
• cross-connects to carriers or partners;
• remote hands support;
• a one-time installation or setup fee.
  
  

As a US retail benchmark, ENCOR Advisors estimated in June 2026 that a 42U full rack with 3 to 5 kW cost about $900 to $2,500 per month at a Tier III facility, depending on the market and contract term. For North American wholesale requirements of 250 to 500 kW, CBRE reported an average asking rate of $196.25 per kW per month in H2 2025. In London, retail colocation runs around $55.73 per rack unit per month (ServerMania, 2026).

These figures are market references, not M247 prices. Actual quotes depend heavily on region and configuration, and the deciding number for many organisations is what leaving an in-house server room saves rather than the monthly rate itself.

Disadvantages of colocation, and when it is not the right fit

The main trade-off is that hardware ownership stays with the customer even though facility operations move to the provider. The organisation still funds server purchases, plans refresh cycles, manages configurations and arranges repairs. That creates a different capital expense profile from public cloud, where the provider owns the underlying equipment.

Physical access also requires a planned site visit or a remote hands request. Migration takes preparation because equipment must be inventoried, transported, installed, connected and tested. Contracts may include fixed terms and committed power levels, which can be less flexible than consuming cloud resources on demand.

Colocation may not fit your organisation if:

• the workload is small, temporary or highly variable;
• you do not want to purchase and maintain hardware;
• you need resources that can scale up or down within minutes;
• your team lacks the skills or service agreements to administer the equipment;
• the hardware is approaching replacement and a different infrastructure model would be more economical;
• the available facility locations cannot meet your latency, compliance or operational requirements.

A workload-by-workload assessment is more useful than choosing one model for the entire organisation. Some systems can remain in colocation while variable applications run in public cloud.

How to choose a colocation provider

Start with the workload, instead of the provider’s feature list. Document the required locations, power density, rack space, connectivity, recovery objectives, access model and expected growth. Then compare providers against the same requirements and ask for evidence behind availability, security and certification claims.

Check the following criteria:

• Location and latency: Choose a facility close enough to users, partners, cloud on-ramps or other infrastructure to meet measurable network requirements.

•  

• Facility design and redundancy: Review the tier claim, power paths, UPS configuration, generators, cooling design and maintenance procedures.

•  

• SLA and operating history: Read the exclusions, service credits, response commitments and reported availability. A design target and actual operating performance are different measures.

•  

• Security and compliance: Verify the scope and current status of relevant certifications, such as ISO/IEC 27001 or SOC 2, and check any sector-specific requirements. A provider certification should cover the facility and services you plan to use.

•  

• Connectivity: Assess carrier choice, cross-connect options, cloud connectivity, route diversity and the cost of adding connections.

•  

• Scalability: Confirm that the site can support the next rack, higher power draw or a private area when required.

•  

• Support: Check access procedures, support hours, escalation paths, remote hands capabilities and response times.

•  

M247 Global offers Tier 3-compliant colocation services with options ranging from single rack units to full racks and custom high-density configurations. Its service includes redundant power and cooling, physical security, cross-connect options, remote hands and 24/7/365 technical support.

Conclusion

The model comes down to owning and managing your servers while renting the facility services that keep them powered, cooled, protected and connected. It can fit SMEs that have outgrown an office server room, enterprises with stable or critical workloads, regulated organisations, network operators and companies expanding into additional regions.

The right choice depends on the workload. Colocation provides hardware control and access to specialised facility infrastructure, on-premises keeps both equipment and facility operations in-house, and public cloud offers provider-owned resources that can be provisioned on demand. Compare performance, compliance, internal skills, hardware lifecycle, cost predictability and scaling requirements before deciding where each workload should run.