There are risks that an organization can address directly. It can secure a building, replicate a server, set up a backup center, install generators, establish emergency procedures, or select alternative suppliers. On the other hand, there are other dependencies that are just as critical but much less controllable: energy, telecommunications, water, transportation, cloud services, data centers, and payment systems. And this is precisely where the most delicate aspect of a resilience manager’s job likely lies. The problem isn’t just knowing that these infrastructures can experience an outage. It’s about understanding how quickly an external outage can spread within the organization and for how long the company is actually able to continue its operations.

The power outage isn’t the real problem

A power outage is a relatively easy scenario to imagine. Well-organized organizations have uninterruptible power supplies (UPS), generators, and emergency procedures in place. In many cases, they have even calculated how long they can operate on their own. However, the truly relevant question is another: How long will this solution continue to function? A generator may have enough fuel for 12, 24, or 48 hours. Once that time has elapsed, however, it must be refueled. At that point, the problem is no longer just technical, but logistical. Does the supplier have sufficient resources? Are the roads passable? Are the gas stations operational? Are there contractual priorities regarding other customers? Is the area affected by the outage the same one on which the fuel supply depends? In other words, continuity does not stop when the emergency system kicks in. Often, it begins at that very moment.

The redundancy may only be apparent

The same applies to telecommunications. An organization may have two connections, two carriers, and multiple communication systems, but that does not necessarily mean it has truly independent infrastructure. Two connections may share the same physical path, the same point of presence, or the same wholesale carrier. Consequently, a seemingly limited outage can simultaneously affect services that, on paper, appeared to be redundant. In a business impact analysis (BIA), the “telecommunications” category is usually listed as a single dependency. In reality, it is a chain of infrastructure, nodes, services, and providers that is rarely mapped out in its entirety. This is precisely where logical redundancy can, in the event of an incident, turn into a single point of physical failure.

The cloud doesn’t eliminate infrastructure—it moves it

Another very common misconception concerns the cloud. Moving applications and data to external infrastructure can reduce the need to directly manage servers, storage, and data centers, but it does not eliminate dependence on infrastructure. It simply shifts it. The organization depends on the cloud provider, telecommunications, authentication systems, DNS, security services, data center power supplies, and a long chain of technological components that are often beyond the company’s direct control. A multi-region or multi-vendor strategy can undoubtedly increase resilience, but only if it is genuinely designed for that purpose. It is necessary to verify that the regions are truly independent, that data is replicated, that applications can be migrated, and that staff are capable of using the backup environment. An alternative does not constitute a continuity strategy simply because it exists on paper. It is only a strategy when it is maintained, verified, and tested.

For banks and other DORA entities, this is not merely a matter of best practice: the regulation requires them to assess the risk of concentration among ICT providers and to define exit strategies for services that support critical or important functions.

The real problem lies in the addictions we don’t see

For a resilience manager, one of the most challenging issues is determining whether all critical dependencies have actually been identified. A function may claim to depend on a specific application. That application depends on a cloud service provider. The cloud service provider depends on specific network infrastructure. Access depends on an identity management system. Emergency communications use the same mobile network that could be affected by the incident. The real dependency is not the first link in the chain. It is the chain in its entirety. This means that a business impact analysis (BIA) may be correct at the first level and yet incomplete. The risk arises when a second- or third-level dependency becomes the true point of failure.

How long does our battery life actually last?

The issue can be summarized in a very concrete question: How long can the organization continue to function without a specific piece of infrastructure? Not without a server or a single vendor, but without electricity, an Internet connection, a mobile network, access to the cloud, payment systems, water, or public transportation. The answer shouldn’t be limited to “we have a plan.” It should be expressed in hours or days. This is the true measure of resilience. An organization that knows what to do but doesn’t know how long it can keep doing it only understands half the problem.

People, too, depend on infrastructure

Furthermore, there is one aspect that is often not given sufficient consideration in plans: people’s availability. During a severe infrastructure crisis, electricity, water, telecommunications, and transportation are not just needed by the organization. They are also needed by employees and their families. A person may be formally available to work but may be unable to get to the workplace, lack an internet connection at home, need to care for family members, or simply be unable to communicate. Therefore, a fully operational emergency center may prove useless if the people who are supposed to use it cannot reach it. This is one of the reasons why infrastructure crises require taking into account resources, technology, people, and the local area all at the same time.

The main risk is the spread

A real-world critical scenario is rarely a simple, isolated failure. Rather, it involves a chain of failures. A power outage can initially be managed using emergency systems. Subsequently, certain telecommunications infrastructures begin to lose their autonomy, connectivity deteriorates, payment systems become intermittent, transportation begins to experience problems, employees struggle to reach their workplaces, and the supply of generators becomes more complex. None of these events, taken individually, is necessarily catastrophic. It is their combination that triggers a crisis. And it is precisely these combinations that are difficult to account for in traditional plans, which are designed for different scenarios.

The 6:30 a.m. phone call

Probably, the most difficult moment for a resilience manager isn’t when they receive news of the incident, but the moment immediately following it. It’s 6:30 a.m. A significant portion of the city is without power, while telecommunications are intermittent. Management calls and asks a seemingly simple question: Can we operate normally? To answer this, information is needed from various departments. The facilities department knows the runtime of the generators, the IT department knows the status of the systems, the procurement department knows the suppliers, the human resources department knows staff availability, and the security department knows the accessibility of the facilities. Business continuity should be the point where all this information is quickly consolidated into a single operational overview. If the first step is figuring out who has this information, the organization is already wasting precious time.

The weak point may be located completely outside the perimeter

An organization may have an up-to-date business impact analysis (BIA), proven plans, backup sites, redundant systems, and qualified vendors. It may have done almost everything right, and yet a single external infrastructure component can simultaneously render many of the implemented solutions unusable. This is where the paradox of modern resilience lies: the more organizations become digital, distributed, and interconnected, the more a significant portion of their operational capacity depends on systems owned by third parties. For this reason, the question that should keep a resilience manager awake at night is not just knowing what could go wrong within the organization, but, above all, which external infrastructures could fail and paralyze it despite everything.