Capacity Planning & What-If Scenario Modeling

VCF 9.0Intermediatevcp-foundation⏱ 105 min

Objectives

Model cluster growth, analyze capacity exhaustion timeline, and recommend infrastructure expansion using what-if scenarios.

Prerequisites

VCF lab environment deployed and operational

Lab Environment

Standard VCF lab environment for Cloud Operations 8.x Professional

Tasks

Task 1 Capacity Planning & What-If Scenario Modeling

Model cluster growth, analyze capacity exhaustion timeline, and recommend infrastructure expansion using what-if scenarios.

Step 1

Analyze Current Capacity State: Run capacity report for prod cluster (custom group)Capture: Total capacity, used, free, growth rate, time remainingResult example: "400GB total, 200GB free, 10GB/week growth → 20 weeks remaining"

Step 2

Scenario 1 - Do Nothing (Baseline): Project capacity exhaustion: 200GB / 10GB/week = 20 weeks (May 2026)Risk: Cluster enters yellow alert (>80% capacity) in 16 weeks (April 2026)Business impact: Procurement lead time is 8 weeks, so deadline to order hardware is February 2026

Step 3

Scenario 2 - Add 2 Hosts (300GB memory total): New capacity: 400GB + 300GB = 700GB totalNew free: 500GB (200GB + 300GB)Time remaining: 500GB / 10GB/week = 50 weeks (April 2027)Cost: 2 hosts × $80K = $160K capitalConclusion: Extends runway to April 2027, meets business planning cycle

Step 4

Scenario 3 - VM Consolidation (reduce demand by 30GB):

Step 5

Scenario 4 - Hybrid Approach (1 Host + Consolidation): Add 1 host (150GB): 400GB + 150GB = 550GBConsolidate 20 VMs: 30GB reductionNew free: 350GB + 30GB = 380GBTime remaining: 380GB / 9.5GB/week = 40 weeks (December 2026)Cost: 1 host ($80K) + 15 hours ops timeConclusion: Optimal balance of capital and effort

Step 6

Present to Finance/Leadership: Create what-if comparison dashboard:Chart: Time remaining (weeks) for each scenarioTable: Scenario name, cost, effort, timeline impactRecommendation: "Scenario 4 - Hybrid (1 host + consolidation) is recommended. Extends capacity to Dec 2026 with $80K investment and 15 hours ops time."

Step 7

Implement Recommended Scenario: Create change request: "Procure 1 host (150GB memory) for prod cluster"Schedule VM consolidation (move 20 VMs to test cluster)Validate post-implementation: Re-run capacity report, verify time remaining matches Scenario 4 projection

Validation Gate

Check: Verify lab completion

Expected: Lab exercise completed successfully

Common Errors

Capacity forecast based on linear growth when actual growth is seasonal

Fix: VCF Operations capacity forecasting uses historical data to project future utilization. If workloads are seasonal (retail: Q4 spike, healthcare: flu season), linear projection underestimates peaks. Use what-if scenarios: model peak periods separately and set procurement triggers based on peak projections, not average.

What-if scenarios not accounting for HA overhead

Fix: A what-if scenario that adds 50 VMs must also account for N+1 HA capacity. If the cluster currently uses 75% capacity with N+1 reserved, adding 50 VMs that consume 10% capacity puts the cluster at 85% — above the 80% recommended threshold for HA + headroom. Include HA overhead in every capacity scenario.

Not setting procurement lead time triggers

Fix: Capacity forecast showing 'will exceed 80% in 90 days' is useless if hardware procurement takes 120 days. Set alert triggers at: current utilization + growth rate × (procurement lead time + buffer). This ensures alerts fire early enough to procure and install hardware before capacity is exhausted.

Final Validation

Lab completed successfully

✓ All steps completed → No errors observed

Cleanup / Restore

• Revert to snapshot if needed

Design Reflection (VCDX)

Capacity planning is a VCDX design component. Panelists test whether your design includes growth projections, procurement triggers, and seasonal capacity modeling.

⚠ Known Pitfalls (from Community KB)

Using linear capacity projections for seasonal workloads — underestimates peak demand.

Setting capacity alerts without accounting for hardware procurement lead time — alerts fire too late to act.

Mark complete