Buying internet connectivity from two different providers can feel like a straightforward way to protect a business against disruption. There are two contracts, two services and two companies to call if something goes wrong. From a procurement perspective, the organisation appears to have removed its dependence on a single supplier.
The underlying network may tell a different story. Connections sold by separate providers can still pass through the same roadside trench, enter a building at the same point or depend on shared backhaul infrastructure. If those common dependencies are not identified, an organisation may have two connections on paper but only one effective route.
Connectivity resilience is a business issue
On 26 September 2026, Ghana’s National Communications Authority announced that voice and data services had been restored across the five Northern Regions and parts of the Oti Region. According to the NCA, the restoration followed repairs to affected fibre infrastructure by the National Information Technology Agency and network operators.
The notice did not provide a detailed account of every technical dependency behind the disruption, so it would be wrong to draw conclusions that the evidence does not support. It does, however, offer a timely reminder of how much economic and everyday activity depends on physical fibre infrastructure.
A network disruption can interrupt more than web browsing and email. Payment processing, customer service, cloud applications, remote access, branch communications and digital public services may all depend on the same underlying connectivity. When those systems are important to operations or revenue, the resilience of the network becomes a business-continuity concern rather than a matter for the IT department alone.
This issue is receiving greater attention in Ghana. Earlier in September, the Minister for Communication, Digital Technology and Innovations called for better coordination to protect fibre infrastructure during road construction and reduce avoidable cuts. Internationally, Uptime Institute’s 2026 outage analysis found that outages linked to fibre and connectivity issues are rising and are more likely to result in extended disruption.
Different providers can still share the same route
Provider diversity and route diversity are often treated as if they mean the same thing. They do not.
Provider diversity means purchasing services from different network operators. Route diversity means ensuring that those services follow sufficiently independent physical and logical paths. An organisation can have the first without necessarily having the second.
For example, two providers may use the same underground duct for part of the journey to a customer site. Their cables may enter the building through the same access point, connect through the same local aggregation facility or depend on the same upstream carrier. Separate physical cables can also share equipment, power or configuration dependencies that allow one incident to affect both services.
The presence of shared infrastructure does not automatically mean that a service is poorly designed. Complete separation may be difficult or commercially unrealistic in some locations. The important point is that customers should understand where dependencies exist, decide whether the resulting risk is acceptable and avoid describing a design as fully redundant without supporting evidence.
What customers should ask their network providers
A provider may not be able to disclose a highly detailed route map because of security or commercial concerns. It should still be possible to provide enough information for the customer to understand the design and identify significant points of common dependency.
When reviewing primary and secondary connections, organisations should ask:
- Do the services follow separate physical routes?
- Do they use different roadside ducts or trenches?
- Do they enter the customer site or data centre through separate access points?
- At what point do the two routes become independent?
- Do the providers use the same backhaul supplier, aggregation site or upstream carrier?
- Do the services share active equipment or power infrastructure?
- Can the secondary connection carry the organisation’s critical traffic?
- Will failover happen automatically, or will someone need to change the configuration?
- How will the providers coordinate if a fault crosses organisational boundaries?
The answers should be reviewed against the needs of the business rather than against a generic definition of resilience. An organisation running a public information website may accept a different level of risk from a bank, payment company, hospital or business whose employees depend on continuous access to central systems.
Connectivity should also be considered as part of the wider service chain. A diverse fibre route will not protect an application if both connections depend on the same misconfigured firewall, router, DNS service or cloud environment. This is why the network assessment should connect with business-continuity planning, application testing and the broader data centre migration process.
A backup connection should be tested
A secondary connection may be installed, monitored and paid for over many months without ever carrying the organisation’s critical traffic. Its presence provides some reassurance, but it does not demonstrate that the business can continue operating when the primary connection becomes unavailable.
A controlled failover exercise can reveal whether traffic moves as expected and whether important applications remain usable. The test should have an agreed window, named decision-makers, defined success criteria and a clear method for returning to the normal configuration. Business representatives should participate where possible because a technically successful failover is of limited value if customers or employees cannot complete essential tasks.
The exercise should record how quickly the interruption was detected, whether traffic moved automatically, which services continued to work and whether the secondary connection had enough capacity. Teams should also check whether security controls, monitoring and incident notifications continued to function during the change.
Testing often exposes problems outside the connection itself. A firewall rule may apply only to the primary route, an external service may recognise only one address, or monitoring may report the failure without confirming that traffic has recovered. Finding these issues during a planned exercise gives the organisation an opportunity to correct them before a genuine disruption occurs.
The role of a carrier-neutral data centre
A carrier-neutral data centre gives customers the freedom to consider services from different connectivity providers. That choice can support a stronger network design, but it does not automatically make two contracted connections physically diverse. The customer and its network providers must still determine how the services are delivered and what they share.
Onix operates a data-centre facility in Accra, Ghana. Onix is a carrier-neutral colocation provider and provides access to multiple networks and internet exchanges. Customers can use that environment to explore connectivity options that reflect their operational requirements, while asking the relevant providers to evidence the routes and dependencies behind each service.
Internet exchanges and local peering can also improve connectivity by allowing networks to exchange traffic more directly. They are valuable parts of the wider connectivity ecosystem, but they answer a different question from physical route diversity. The Onix article What is Peering? explains the distinction between peering, internet transit and local traffic exchange.
Understand the dependencies before an outage reveals them
Purchasing a second internet connection is a sensible step, but the number of connections alone does not establish resilience. Organisations need to understand where the services run, which elements they share, how failover is expected to work and whether that process has been tested with the systems the business actually uses.
This does not require every business leader to become a network engineer. It requires the technical, procurement and business-continuity teams to ask more precise questions and insist on evidence that matches the importance of the service.
If your organisation is assessing colocation and connectivity options in Accra, speak with Onix about the options available at the facility. As part of that discussion, ask your network providers to explain how their proposed services address end-to-end route diversity and any remaining shared points of failure.
Frequently asked questions
Do two internet providers guarantee network redundancy?
No. Connections from separate providers may still share ducts, building entry points, backhaul infrastructure, equipment or upstream services. Organisations should establish which physical and logical dependencies remain common to both services.
What is physical route diversity?
Physical route diversity means connections follow sufficiently independent physical paths. Depending on the location and the organisation’s requirements, this may include separate roadside ducts, trenches and building entry points.
How often should network failover be tested?
There is no single timetable that applies to every organisation. The frequency should reflect the importance of the connected services, regulatory or contractual requirements, previous test findings and the frequency of changes to the network. Testing should also be considered after significant provider, equipment or configuration changes.
Is carrier neutrality the same as route diversity?
No. Carrier neutrality gives customers a choice of network providers. Route diversity depends on how the selected connections are designed and delivered.
