Lab B1: Deploy Avi Controller HA Cluster & Service Engine Fabric in VCF 9.0

VCF 9.0Intermediatespecialist⏱ 75 min

Objectives

Stand up a production-style Avi deployment with 3-node Controller cluster and SE fleet on VCF 9.0 NSX-T cloud.

Prerequisites

VCF lab environment deployed and operational

Lab Environment

Standard VCF lab environment for Avi Load Balancer 30.x Specialist

Tasks

Task 1 Lab B1: Deploy Avi Controller HA Cluster & Service Engine Fabric in VCF 9.0

Stand up a production-style Avi deployment with 3-node Controller cluster and SE fleet on VCF 9.0 NSX-T cloud.

Step 1

Deploy Avi Controller 30.x as a 3-node cluster in management workload domain.

Step 2

Integrate Avi with NSX-T cloud: connect to NSX Manager, import transport zones.

Step 3

Create SE groups for elastic HA (N+M buffer) and dedicated tenant isolation.

Step 4

Deploy first 2 SEs and verify auto-scale to 4 under simulated load.

Step 5

Validate Controller-SE heartbeat, cert bootstrapping, and control plane health.

Validation Gate

Check: Verify lab completion

Expected: Lab exercise completed successfully

Common Errors

Avi Controller cluster with fewer than 3 nodes

Fix: Avi Controller cluster requires 3 nodes for HA with quorum. 2-node cluster has no quorum — a single node failure makes the cluster read-only. Single-node is acceptable for labs only. Production requires 3-node cluster with anti-affinity rules (separate ESXi hosts).

Service Engine group not scoped to correct VCF networks

Fix: Service Engines must connect to both front-end (client) and back-end (server) networks. If the SE group is configured with the wrong management or data network, SEs deploy but cannot reach pool members (backend servers). Verify SE network placement matches the application topology.

SE VM resource allocation too small for throughput requirements

Fix: SE performance scales with allocated vCPU and memory. Default SE (2 vCPU, 2GB RAM) handles approximately 5 Gbps. For higher throughput, scale SE resources or deploy multiple SEs with ECMP. Size based on: expected peak throughput + 50% headroom.

Final Validation

Lab completed successfully

✓ All steps completed → No errors observed

Cleanup / Restore

• Revert to snapshot if needed

Design Reflection (VCDX)

Avi deployment architecture shows application delivery design skills. VCDX panelists test SE placement, Controller HA, and throughput sizing methodology.

⚠ Known Pitfalls (from Community KB)

Deploying a 2-node Controller cluster — no quorum means single-node failure makes the cluster read-only.

Undersizing Service Engines — default 2 vCPU SEs are insufficient for production throughput.

Mark complete