Changing Engines in Mid-Flight: Live Upgrade from NSX-V to NSX-T
Upgrading core overlay SDN infrastructure while processing live production transactions. How we migrated from NSX-V to NSX-T with zero downtime.
βReplacing your software-defined networking control plane while thousands of production VMs are actively communicating is like swapping a jet engine mid-flight.β
During my tenure as a VMware NSX Consultant at IBM, we executed live SDN migrations for enterprise clients transitioning from legacy NSX-V (vCenter-bound) to NSX-T (multi-hypervisor, container-ready).
The Control Plane Disconnect
- NSX-V: Dependent on VMware vCenter Server, uses VXLAN encapsulation, single-vCenter scope.
- NSX-T: Autonomous NSX Manager cluster, uses GENEVE encapsulation, supports KVM, ESXi, and bare metal.
[!IMPORTANT] Because NSX-V uses VXLAN (port 4789) and NSX-T uses GENEVE (port 6081), hosts cannot communicate across overlays without explicit L2 Bridging or Tier-0 routing cutovers.
# # Verify NSX-T Transport Node Tunnel Endpoint (TPEP) Status
# esxcli network ip interface ipv4 get -i vmk10
Interface IPv4 Address IPv4 Netmask Type Gateway
vmk10 192.168.100.50 255.255.255.0 STATIC 0.0.0.0 # GENEVE TTEP
The Verdict
Key Takeaway
Isolate Control Plane Migration from Data Plane Cutover.
Migrate overlay transport nodes in maintenance batches while maintaining dual-stack edge routing. Software-defined network migrations require strict separation of control plane steps from data plane cutovers.
Sachin Kumar Sharma
Associate Director (Infrastructure & Cloud Architecture Strategy) | 20+ Yrs Exp
Architecting resilient multi-cloud enterprise landing zones, SDN overlay fabrics, DevSecFinOps automation pipelines, and autonomous Agentic AI platforms.