Colocation Contract: Uptime, Power and Exit Rights

Before signing a colocation contract, check how uptime is defined and measured, what power capacity and redundancy are actually promised, and whether service credits are your only remedy for outages. Also pin down renewal, termination, equipment removal, migration assistance, and fees in writing; marketing claims about availability do not replace contract language.

How uptime commitments work

Colocation providers supply space and supporting infrastructure, while the customer typically owns and operates the servers and applications. A contract may promise availability for a particular facility service, such as conditioned space or electrical service to a defined handoff point. That is not necessarily a promise that every server, network path, or application will be available.

Read the service-level agreement (SLA) alongside the main agreement, order form, and any service descriptions. Confirm that the documents use the same service names and that the SLA applies to the specific site, cabinet, cage, or capacity being purchased. Check which document prevails if terms conflict.

• Availability definition: Identify precisely what counts as an outage: loss of power at the agreed demarcation point, loss of cooling, inability to access the facility, or another measurable event.
• Measurement period: Check whether availability is calculated monthly or over another period, and how partial outages and intermittent failures are treated.
• Exclusions: Review planned maintenance, customer-caused incidents, upstream utility failures, force majeure, and other exclusions. Broad exclusions can make a headline percentage less meaningful.
• Evidence and notice: Find out who detects an incident, what records are used, how quickly the provider must notify you, and whether you must submit a claim within a deadline.

A stated availability percentage is a service commitment, not proof that the customer’s entire IT service will meet that level. For example, a facility may deliver power as promised while a customer’s single server fails. A useful agreement distinguishes facility responsibility from customer equipment, cabling, network services, and applications.

Power capacity, redundancy and limits

Verify the electrical commitment in engineering terms, not just a cabinet’s advertised size. The order should state the contracted capacity, how it is measured, the applicable voltage and connection type, and any limits on continuous draw, peak draw, or power factor. Confirm whether a quoted figure is per cabinet, per feed, or shared across a larger allocation.

Ask what “A and B power” means at the actual delivery point. Two feeds can provide separate paths, but their independence depends on the design and on how far that separation extends. Determine whether each feed can support the full load if the other fails, and whether your equipment has appropriately rated dual power supplies. A redundant facility feed cannot protect a device connected to only one feed.

• Specify the number and type of circuits, connectors, and metering provided, plus who supplies and maintains downstream whips or cabling.
• Ask how capacity is monitored, how overages are handled, and whether the provider can throttle or disconnect equipment—and under what notice and emergency procedures.
• Check whether power is billed by reserved capacity, measured consumption, or another stated method, and how readings and disputes are handled.
• Review the process and lead time for increasing or reducing capacity; availability should not be assumed merely because unused capacity appears to exist.

Cooling, generator fuel arrangements, transfer systems, maintenance, and utility interruptions may also affect continuity. The contract should make clear which of these are covered by an SLA and which are described only as operating practices. Do not assume that a redundancy label such as “N+1” guarantees a specific end-to-end result without a definition of its scope and the relevant failure scenarios.

Outage remedies and incident handling

Service credits are common SLA remedies, but their value depends on the calculation and restrictions. Check the credit tiers, the amount to which a percentage applies, any monthly cap, whether credits apply automatically, and whether they can be used only against future invoices. A credit based on a small recurring charge may be modest compared with the consequences of downtime.

Read whether credits are the exclusive remedy for an SLA failure, and how that interacts with termination rights, liability limits, indemnities, and any separately negotiated remedies. These provisions are contract-specific; a service credit does not automatically establish compensation for every loss. Have counsel review language that limits remedies or excludes categories of damages.

Incident procedures matter operationally. Look for a support contact available during outages, escalation levels, response and update commitments, access to incident reports, and cooperation with root-cause analysis. A response-time target is not the same as a restoration-time guarantee. Ask whether the provider must preserve relevant logs and explain how recurring incidents are escalated.

Term, renewal and price changes

Map the initial term, renewal dates, notice windows, and the method for giving notice. Automatic renewal can create an unwanted extra term if notice must be sent months in advance or to a particular address. Confirm when recurring charges begin, how installation or cross-connect charges are treated, and what happens to fees if capacity is added or removed during the term.

Check the provider’s rights to change prices, policies, service descriptions, or facility procedures. If the provider can make material changes unilaterally, ask for advance notice and a right to terminate without penalty when a change materially affects the service. Distinguish ordinary operational changes from changes to core capacity, availability, access, or charges.

Exit rights and equipment removal

Exit terms can determine whether a technically successful migration becomes an expensive or rushed one. Identify termination for convenience, termination for provider breach, any cure period, and the consequences of a prolonged or repeated outage. Check whether notice must be given before the end of the term and whether early termination charges apply. If a charge is tied to remaining contract fees, understand its calculation and whether it decreases over time.

Plan the physical exit as carefully as the legal one. The contract should explain how access is scheduled, who may remove equipment, what identification or authorization is required, and whether packing, loading, or after-hours work costs extra. Confirm the deadline for removing hardware after termination and the consequences if equipment remains. Address data handling too: specify responsibilities for wiping or destroying storage media, documentation of the process, and any fees. Do not assume the provider will securely erase customer-owned drives unless the agreement says so.

• Check whether the provider must provide reasonable migration cooperation, records, or access during a transition, and whether that help is separately charged.
• Confirm how prepaid amounts, deposits, disputed invoices, and final meter readings are reconciled.
• Ask how the agreement treats an insolvency, facility closure, material service change, or a transfer of the provider’s business.
• Make sure termination of one service or location does not unintentionally terminate—or leave you paying for—other orders.

Keep copies of the signed order, SLA, site rules, and amendments. For multi-site arrangements, check that each location has its own applicable service terms and exit process rather than assuming one site’s commitments apply everywhere.

Frequently Asked Questions

Does a 99.99% uptime SLA guarantee my servers will be available?
No. It generally covers only the service and measurement point stated in the SLA. Customer hardware, operating systems, network paths, and applications may be outside that commitment.

Are service credits the same as compensation for an outage?
Not necessarily. Credits are contractual billing adjustments, often limited by caps and claim rules. Check whether they are the exclusive remedy and how other liability provisions apply.

What does A/B power mean in a colocation contract?
It usually refers to two power feeds, but the label alone does not establish their independence or capacity. Confirm their routing, delivery point, and whether either feed can carry the full required load.

Can I leave a colocation contract early?
That depends on termination provisions, the term, breach rights, notice rules, and any early termination charges. Do not rely on a sales discussion where the signed documents say otherwise.

What happens to equipment when a colocation contract ends?
The contract should set the removal deadline, access procedure, charges, and treatment of equipment left behind. It should also clarify responsibility for data on storage devices.

Should the SLA cover cooling as well as power?
Check whether cooling is part of the defined service and whether it has a separate availability measure or remedy. Power commitments alone do not necessarily address temperature or cooling interruptions.

Related

• Compare Refrigerator Energy Use & Running Costs
• Apple Watch Remote iTunes: Quick Guide
• Best Alternatives to Optical Drives