Academy/VCAP — VCF Networking (3V0-25.25)/OSPF on a Tier-0 Gateway & Route Redistribution
This lab targets VCF 9.0

OSPF on a Tier-0 Gateway & Route Redistribution

VCF 9.0Intermediatevcap-advanced⏱ 150 min

Migrated from v1 Advanced VCF 9.0 Networking; rebuilt in v8.2.2 around the documented NSX 9.0 OSPF capabilities (single area per Tier-0, OSPFv2, default VRF, no redistribution between BGP and OSPF)

Objectives

  • Enable OSPFv2 on a Tier-0 gateway with a single area (normal or NSSA) peering with the physical/virtual router
  • Redistribute Tier-0 and Tier-1 networks and a static default route into OSPF and read the resulting external LSAs
  • Verify adjacencies and routes from the Edge CLI and from the peer router; back the session with BFD
  • State the NSX OSPF restrictions that shape the design (single area per Tier-0, default VRF only, no BGP↔OSPF redistribution)

Prerequisites

VCF 9.0 workload domain with an NSX Edge cluster (2 Edge nodes) and a Tier-0 gateway 'T0-OSPF' with two external interfaces per Edge node on an uplink VLAN; a physical or virtual router (VyOS in Holodeck) on the same VLAN running OSPF area 0 and holding a default route to the 'ISP'; a Tier-1 'T1-App' with segment 10.100.0.0/24 and a test VM

Prior labs: net-virt-02

Required skills:

  • Tier-0/Tier-1 routing and route redistribution
  • OSPF fundamentals (areas, LSA types, external routes)
  • NSX Edge CLI

Lab Environment

One Tier-0 gateway (Active-Active, ECMP) on edge-node-01/02, uplinks 192.168.100.11/.12 (/24, broadcast) towards router R1 192.168.100.254. R1 is in OSPF area 0 and advertises external networks; it holds a static default route 0.0.0.0/0 → 10.0.0.1 (ISP). Tier-1 'T1-App' with segment 10.100.0.0/24 hangs off the Tier-0.

Tasks

Task 1 OSPF on a Tier-0 Gateway & Route Redistribution

NSX OSPF (OSPFv2 only) runs on the Tier-0 gateway in the default VRF, Active-Active or Active-Standby, with exactly one area definition per Tier-0 (backbone, normal or NSSA; there is no stub type) and a maximum of two Tier-0 uplinks per Edge node. The Tier-0 is therefore never an ABR: inter-area design lives on the physical network. Redistribution mirrors BGP (Tier-0 uplinks/downlinks/loopbacks, Tier-1 downlinks, static) and produces Type 5 external LSAs in a normal area or Type 7 in an NSSA, type-2 metric only; redistribution route maps match prefix lists only (no set actions). No redistribution is supported between BGP and OSPF, so OSPF-to-BGP loops cannot arise on the Tier-0 — the two protocols coexist only for BGP multi-hop peers learned via OSPF. ECMP up to 8 paths, Default Originate, BFD per interface, Helper-only graceful restart, MD5/plain authentication; NAT and Tier-0 VIP are not supported with OSPF, and OSPF is not supported in Federation.

Replace the BGP peering of Lab net-virt-02 with OSPF on the Tier-0, redistribute the overlay networks and a default route, and verify the exchange from both ends — within the documented NSX limits.

Step 1

Prepare the Tier-0. Networking → Tier-0 Gateways → T0-OSPF → Edit. Confirm HA Mode Active-Active with edge-cluster-01, and under Interfaces two external interfaces per Edge node on the uplink VLAN segment (192.168.100.11 and .12 /24; a third uplink per node is not allowed with OSPF). No NAT and no Tier-0 VIP on this gateway (unsupported with OSPF). Save.

Tier-0 with 2 uplinks per Edge node, no NAT, no VIP.
Step 2

Enable OSPF with one area. Tier-0 → Routing (OSPF section) → OSPF toggle On → Area Definition → Set → Add Area Definition: Area ID 0 (a number or IPv4-formatted value), Type Normal, Authentication MD5 with a key ID and password shared with R1 → Save. Graceful Restart: Helper Only. ECMP: On. Observe that the UI allows only one area definition.

Area 0 defined; attempt to add a second area is not offered.
Step 3

Attach the uplinks to OSPF. OSPF Configured Interfaces → Set → Configure Interface for each of the four external interfaces: Area ID 0, Network Type Broadcast (P2P is for /31 links), OSPF Enabled, BFD Enabled with the default profile (500 ms, the Edge VM minimum), Hello 10 s / Dead 40 s (defaults) → Save. Click Save on the gateway.

Four OSPF interfaces listed in area 0 with BFD.
Step 4

Configure R1 (VyOS): set protocols ospf parameters router-id 192.168.100.254; set protocols ospf area 0 network 192.168.100.0/24; set protocols ospf area 0 authentication md5 (matching key); set protocols ospf redistribute static; set protocols static route 0.0.0.0/0 next-hop 10.0.0.1; commit. On each Edge node (admin CLI), enter the Tier-0 SR VRF first (get logical-routers; vrf <id> — VCF 9.0.1 docs p.2784), then: get ospf neighbor — R1 appears in state Full from both nodes; get ospf interface lists the uplinks with their DR/BDR roles and Hello/Dead timers (NSX CLI Guide: get ospf neighbor, get ospf interface).

Two Full adjacencies (one per Edge node) with R1.
Step 5

Redistribute the overlay networks. Tier-0 → Route Re-distribution → OSPF Route Redistribution Status On → Set → Add rule 'ospf-redist': Destination Protocol OSPF, sources Tier-0 Connected Interfaces (uplinks excluded if desired), Tier-1 Connected Segments, Tier-1 NAT IPs; optionally attach a route map that matches a prefix list (set actions are not supported). On T1-App ensure Route Advertisement → Connected Segments is on. Save.

On R1: show ip route ospf shows 10.100.0.0/24 as an O E2 (type-2 external) route via both Edge uplinks (ECMP).
Step 6

Read the LSAs. On an Edge node, in the Tier-0 SR VRF: get ospf database (NSX CLI Guide) shows Router (Type 1) and Network (Type 2) LSAs for area 0 and AS-External (Type 5) LSAs for the redistributed prefixes; there are no Summary (Type 3) LSAs because the Tier-0 is in a single area and is not an ABR. Change the area type to NSSA (R1 too) and observe the externals become Type 7 LSAs; change back to Normal.

Type 1/2/5 LSAs in Normal; Type 7 in NSSA; no Type 3.
Step 7

Learn the default route. R1 redistributes its static default (set protocols ospf redistribute static, or set protocols ospf default-information originate always). On the Edge: get route 0.0.0.0/0 shows the OSPF-learned default via R1 (o* route) on both nodes. Alternative direction for a Tier-0 that owns the default: the Default Route Originate toggle in the OSPF section advertises 0.0.0.0/0 into OSPF.

Default route present on both Edge nodes via R1.
Step 8

End-to-end test and failure. From the test VM on 10.100.0.0/24: ping 10.0.0.1 (ISP side via R1). Then shut one Edge node's uplink on R1 (or disable the interface on the Edge): BFD (500 ms x 3) detects the loss in about 1.5 s and traffic continues over the second Edge node; get ospf neighbor on the surviving node still shows Full. Restore.

Continuous reachability through the remaining Edge; adjacency count 2 → 1 → 2.
Step 9

Route summarization and cost. Tier-0 → OSPF → Route Summarization → Set → Add Prefix 10.100.0.0/16 Advertise Yes — R1 now sees a single summary external for the Tier-1 segments. Note that OSPF interface cost is not an NSX per-interface setting here; path preference is driven on R1 (or by ECMP on the Tier-0). Record both decisions.

R1 shows the 10.100.0.0/16 summary instead of individual /24s.
Step 10

Record the restrictions as design decisions: one area per Tier-0 (inter-area design belongs to the physical network), default VRF only, no BGP↔OSPF redistribution (OSPF can only learn the BGP peer address for multi-hop BGP), Helper-only graceful restart, OSPFv2 only (IPv6 routing needs BGP or static), no OSPF in Federation, NAT and Tier-0 VIP unsupported.

Decision list with justification and implications.

Validation Gate

Check: Single-area OSPF on the Tier-0 exchanging routes with R1

Expected: Two Full adjacencies with BFD; 10.100.0.0/24 (or its /16 summary) learned on R1 as a type-2 external via ECMP; default route learned on both Edge nodes; Type 5/7 LSAs observed, no Type 3; traffic survives one Edge uplink failure

Common Errors

Adjacency stuck in Init/ExStart or not forming
Fix: Check area ID and type match (Normal vs NSSA; NSX has no stub type), MD5 key ID/password, network type (Broadcast vs P2P), matching Hello/Dead timers (NSX defaults 10/40 s) and the same subnet mask: 'get ospf neighbor' on the Edge, 'show ip ospf neighbor' on R1.
Redistributed prefixes missing on R1
Fix: OSPF Route Redistribution Status toggle is off, the rule's sources do not include Tier-1 Connected Segments, Tier-1 Route Advertisement is off, or a route map prefix list does not match. Remember redistribution produces type-2 externals (E2/N2) only.
Expecting OSPF-learned routes in BGP (or vice versa)
Fix: No redistribution is supported between BGP and OSPF on the Tier-0; design the exchange on the physical router if both protocols are needed, or use OSPF only to reach a multi-hop BGP peer.
OSPF hello/dead timer mismatch causing flapping adjacencies
Fix: OSPF adjacency requires matching hello and dead intervals. NSX default: hello 10 s, dead 40 s, set per OSPF interface. Match timers on both sides and rely on BFD for fast failure detection.

Final Validation

OSPF single-area peering and redistribution operational on the Tier-0

✓ Adjacencies → get ospf neighbor shows R1 Full on both Edge nodes; BFD up

✓ Redistribution → R1 learns Tier-1 segment(s) as O E2 (or N2) routes via two next hops

✓ Default route → get route 0.0.0.0/0 shows the OSPF-learned default on both nodes

✓ LSAs → Type 1, 2 and 5 (or 7 in NSSA) present; no Type 3

✓ Failure → Traffic continues through the surviving Edge after an uplink failure

Cleanup / Restore

• Disable OSPF on T0-OSPF and remove the area definition

• Remove the OSPF configuration from R1 (VyOS: delete protocols ospf; commit)

• Re-enable the BGP peering from Lab net-virt-02 if the environment depends on it

Design Reflection (VCDX)

OSPF on a Tier-0 is a constrained design choice and the panel will test whether you know the constraints: single area, default VRF, OSPFv2 only, type-2 externals, no BGP↔OSPF redistribution, Helper-only GR, no Federation. Defend OSPF only where the customer's physical network is OSPF-based and BGP is unavailable; otherwise BGP remains the VCF default. Show the loop-prevention story (no redistribution on the Tier-0, prefix-list route maps) and the convergence story (BFD 500 ms x 3 on Edge VMs, ECMP across Edge nodes).

Requirements

  • Dynamic routing between the Tier-0 and an OSPF-only physical network
  • Fast failover between Edge nodes (BFD)

Constraints

  • One OSPF area per Tier-0; max two uplinks per Edge node
  • Default VRF only; no NAT or Tier-0 VIP with OSPF; no redistribution between BGP and OSPF

Assumptions

  • The physical router supports OSPFv2 with MD5 authentication and BFD

Risks

  • Inter-area design and summarisation errors on the physical network propagate into NSX as external routes

Self-Assessment Discussion Prompts

  1. When would you choose OSPF over BGP for the Tier-0 uplinks in a VCF design, and what do you give up?
  2. How does the single-area restriction change a multi-site routing design?

Extensions

Replace the Normal area by an NSSA on both ends and compare the LSA database

Add a loopback interface to the Tier-0 and observe the router-ID rules (highest interface IP unless the loopback is chosen after an OSPF restart)

Combine OSPF (to learn the peer address) with multi-hop BGP to the same router

⚠ Known Pitfalls

Designing the Tier-0 as an OSPF ABR or with several areas — NSX allows one area definition per Tier-0
Planning BGP↔OSPF redistribution on the Tier-0 — not supported; the physical network must do it
Mismatching OSPF timers or area type between NSX and the physical router — adjacency never completes or flaps
Enabling NAT or a Tier-0 VIP on an OSPF Tier-0 — both unsupported with OSPF

References

  • VCF 9.0 documentation — Configure OSPF on a Tier-0 Gateway (capabilities and restrictions; area definition; OSPF Configured Interfaces; route redistribution): https://techdocs.broadcom.com/us/en/vmware-cis/vcf/vcf-9-0-and-later/9-0.html
  • VCF 9.0 documentation — Add an NSX BFD Profile (500 ms minimum for Edge VMs): https://techdocs.broadcom.com/us/en/vmware-cis/vcf/vcf-9-0-and-later/9-0.html
Was this page useful?
Type to search. ↑ ↓ to move, Enter to open, Esc to close.