𝄜 OSPF vs. BGP for SD-WAN | Hub-and-Spoke vs. Mesh ⿻

When designing an SD-WAN underlay, choosing between OSPF and BGP depends on the architecture:

Let’s break this down.

1. Hub-and-Spoke Architecture

✅ Pros:

Scales well – Hubs can act as route reflectors to simplify peering.
No flooding issues – Unlike OSPF, BGP does not flood link-state changes.
Better control – BGP policies can be used to influence routing paths.
Supports multi-homed setups – Ideal for SD-WAN scenarios where the hub has multiple upstream ISPs.

❌ Cons:

Slower convergence – BGP is not designed for real-time failover. SD-WAN overlay mechanisms (like path monitoring) are needed.
More complex configuration – Requires route reflectors or confederations for scalability.

❌ OSPF has scalability issues in hub-and-spoke.

Every spoke learns every route from the hub. In large deployments, this can lead to LSA (Link-State Advertisement) overload.
Does not scale well across different WAN links.

➡ Verdict: BGP is the best choice for SD-WAN hub-and-spoke deployments.

OSPF is problematic because of excessive flooding in a large mesh.
BGP scales better but has slow convergence (not good for fast SD-WAN failover).
SD-WAN’s own overlay intelligence (policy-based routing, dynamic path selection) usually replaces the need for full-mesh IGP routing.

Babel (or EIGRP in Cisco environments) handles dynamic networks better than BGP/OSPF, but it's rarely seen in SD-WAN.
SD-WAN solutions like Fusion use their own intelligent path selection, which outperforms any IGP in a full-mesh SD-WAN setup.

➡ Verdict: Avoid full mesh unless necessary. If you must use it, use BGP over OSPF, but rely on SD-WAN's native traffic steering instead.