Most migration plans focus on one question: which Lagos data centre do we move to? That is the wrong question. The right question is: what is our onshore disaster recovery topology?
Here is the data. Nigeria has about 28 data centre facilities. Roughly 18 are commercial. The combined live capacity across all of them is about 50 to 56 megawatts. Including expansion projects already under construction, installed capacity rises to about 124 MW.
About 84% of that capacity is in Lagos. The facilities sit in Lekki, Ikeja and Eko Atlantic. The rest of the country shares the remaining 16%. Abuja has some. Kano has Galaxy Backbone's Tier IV facility. Port Harcourt has a new Equinix site. Enugu has a small presence.
This concentration is a problem. If you move payment transaction data to a Lagos colocation and Lagos has a major outage, your compliance is intact (the data is onshore) but your operations are not (the data is inaccessible). The CBN does not care which data centre the data lives in as long as it is in Nigeria. Your customers care whether the service is running.
A Lagos-only onshore strategy replicates the single-region risk that the hyperscalers eliminated years ago. When a bank ran on AWS af-south-1 (Cape Town) with a DR plan in eu-west-2 (London), it had geographic diversity. If Cape Town went down, London took over. Moving to a single Lagos colocation with no second site is not a resilience improvement. It is a regression.
What the operators say
The data centre operators acknowledge the capacity question but argue it is manageable. OADC's chief executive, Ayotunde Coker, said in a press briefing that "there is no capacity problem in the high-quality data centres with expansion plans already in place." He pointed to OADC's 24 MW expansion in Lekki, and to similar builds by Equinix, Rack Centre and Airtel's Nxtra.
Kasi Cloud's chief executive, Johnson Agogbua, was more measured. He said physical capacity is not the bottleneck. The real constraint, he argued, is whether Nigeria has enough locally available cloud computing and storage platforms to support the migration at scale. The data centres provide power, cooling and connectivity. They do not provide the cloud platform layer that banks and fintechs actually consume.
On the concentration risk, the operators are quieter. Nobody wants to say it is a problem if all the data goes to one city. But the math is the math. If most of the live capacity is in Lagos and most institutions migrate to Lagos, the outage surface grows, not shrinks.
Why concentration matters more now
Single-region risk was always a theoretical concern. For most Nigerian banks, the production estate was on AWS or Azure with multi-region failover built in by the cloud provider. The risk was managed by someone else's architecture.
After the migration, that changes. The onshore estate will not have multi-region failover in the same sense. You can replicate between two Lagos facilities, but a Lagos-wide event (power grid failure, fibre cut, flooding) takes out both. You need geographic diversity.
Galaxy Backbone's Kano facility is the strongest option for a second site. It is Tier IV certified (the highest tier for fault tolerance), PCI-DSS certified for card data, and connected to GBB's national fibre network. It was designed for disaster recovery and business continuity for government systems. It can serve the same function for financial institutions.
The problem is adoption. Banks and fintechs have been slow to commit to Kano as a DR site. The perception is that Kano is far from the commercial centre, that talent is harder to find there, and that the GBB facility is designed for government workloads, not commercial financial services. Some of those perceptions are outdated. GBB has been actively courting bank chief information officers through second-quarter webinars. The infrastructure exists.
Other options exist at smaller scale. Equinix's Port Harcourt facility is Tier III. Abuja has some colocation capacity. None of these match Kano's certification profile or GBB's fibre connectivity.
What good looks like
A resilient onshore topology has three layers.
Primary onshore. The main Lagos colocation. This runs the "must be onshore" workloads with synchronous replication to the DR site. This is where the transaction data lives day to day.
Secondary onshore. The DR site in Kano, Abuja or Port Harcourt. This runs the same workloads with near-synchronous replication from the primary site. If Lagos goes down, this site takes over within minutes. The DR site should be able to run the full payment transaction load independently, not just a subset.
Offshore. The existing hyperscaler region. This runs the "can stay hybrid" and "should stay offshore" workloads (analytics, AI training, archival). It also serves as a third-layer DR option for non-transaction workloads. If both onshore sites go down (unlikely but possible), the offshore environment keeps the business running for everything not subject to the circular.
Why this is a board-level discussion
A Lagos-only migration is cheaper and faster. The design is simpler. The contracts are easier. The operational overhead is lower. A two-site topology costs more, takes longer and requires more coordination.
But the decision is not just an architecture decision. It is a risk decision.
If a bank migrates all payment transaction data to a single Lagos colocation and Lagos has a multi-day outage, the bank cannot process payments. That is not a technical failure. It is a board-level failure of resilience planning. The board needs to decide whether the cost of a second site is worth avoiding that scenario.
The CBN has already signalled its enforcement posture through the POS geotagging mandate and the data localisation circular. Resilience requirements are a reasonable next step. A bank that can demonstrate a two-site onshore topology is in a stronger position than one that cannot.
What to do this month
If you are planning a data localisation migration, add these items to the roadmap:
- Evaluate second-site options now, not after signing a Lagos colocation contract. The contract terms will be better if primary and DR are bundled.
- Test the latency between the Lagos colocation and the DR site. Near-synchronous replication at financial transaction volumes requires predictable latency. Find out what it is before designing the replication topology.
- Design the application layer to support multi-site failover. If the application assumes a single data centre, the migration is a good time to make it site-agnostic.
The January 1 deadline is driving most migration decisions. But the migration is not the end. The end is a resilient, compliant and operable onshore infrastructure. That takes more than one data centre in one city.
The position of this Journal
A two-site topology is not a luxury add-on. It is the difference between an onshore estate that survives and one that concentrates the country's payment record in a single point of failure. Kaliabe designs the full topology: the Lagos primary, the second-site DR, the offshore boundary, and the replication paths between them. Institutions that treat resilience as a compliance checkbox will find out what the checkbox was worth on the day Lagos goes dark.
FIG. J5 — ONSHORE TOPOLOGY · LAGOS PRIMARY · KANO DR · OFFSHORE · THREE LAYERS, ONE OUTAGE SURFACE