Across the Nigerian banking sector, particularly tier-2 commercial banks, the larger microfinance institutions, and the fast-growing fintech segment, we are seeing a quiet but unmistakable shift away from traditional MPLS toward SD-WAN. The economics have flipped. MPLS contracts that used to feel like the safe choice now feel rigid, expensive, and harder to scale to a branch network that's still adding sites every quarter.
Why the conversation is now happening
- Branch counts keep growing while MPLS pricing per site keeps creeping up.
- ATM and POS uptime has become a customer-experience metric, not just an operations one.
- CBN's Risk-Based Cybersecurity Framework expectations on segmentation, encryption, and resilience map cleanly onto modern SD-WAN architectures, and badly onto flat MPLS designs.
- Starlink's arrival gives every branch a credible third underlay for the first time.
- Cloud-hosted core banking and fraud-analytics traffic doesn't behave the way MPLS was designed for.
Where banks get the migration wrong
- Treating SD-WAN as a like-for-like replacement for MPLS instead of a rearchitecture, and ending up paying for both during a long, expensive parallel run.
- Migrating without a proper segmentation design, the cardholder data zone needs to be isolated more rigorously after the migration, not less.
- Ignoring the fact that ATM switch and payment-processor third-party tunnels need explicit re-engineering.
- Underestimating how much of the value comes from the centralized monitoring, not the bandwidth savings.
- Picking an SD-WAN platform without a clear story for CBN cyber-resilience compliance.
Our default migration playbook for FSI
- Pilot at 3–5 representative branches (one HQ-adjacent, one urban, one semi-urban, one rural) to validate failover and SLA behaviour against real production load.
- Run MPLS and SD-WAN in parallel during pilot, measure, don't argue.
- Design segmentation up front: cardholder data, branch operations, ATM, third-party, guest, all distinct from day one.
- Bond Starlink as a third underlay where it materially improves resilience, not as a primary cost-cutter.
- Move the management plane to the cloud (with a Nigerian-compliant gateway) before the data plane, visibility is what makes the rest of the migration safe.
- Wave-out MPLS site-by-site with clear rollback gates. Don't big-bang.
Done well, the migration delivers 15–25% bandwidth-cost reduction at scale, dramatically better ATM/POS uptime, and a CBN-compliance posture that's easier to audit. Done poorly, it delivers a paper-saving that gets eaten by an outage. The architecture conversation is more important than the contract.