Academy/VCDX Defense Preparation
VCF

VCDX Defense Preparation

VCF 9.0vcdx-distinguished

Master the VCDX (VMware Certified Distinguished Expert) certification path, design document structure, RCAR framework, defense strategy, and common failure scenarios. Comprehensive guidance for architect-level professionals pursuing VMware's highest certification.

Version Evolution

The VCDX certification has evolved from vSphere-only design expertise to the full VCF stack. On 13 Nov 2025 it was re-launched as the VMware Certified Distinguished Expert with Architect, Administrator and Support tracks, each built on a VCP-VCF, three domain VCAP-VCF exams (VCAP-VKS mandatory) and the matching role-based VCAP-VCF. Modern VCDX candidates must demonstrate proficiency across compute (vSphere), storage (vSAN), networking (NSX), automation (VCF Automation), operations (VCF Operations) and Kubernetes (VKS). Defence format: the legacy 2024 guide describes a 120-minute defence (a 75-minute oral defence followed by a 45-minute design problem); the format of the new VCDX-VCF Architect panel has not been published as of 24 Sep 2026.

Learning Outcomes

  • Explain the 2026 VCDX tracks and their prerequisite chain: VCP-VCF, three domain VCAP-VCF exams (VCAP-VKS mandatory) and the matching role-based VCAP-VCF.
  • Document every design trade-off with its justification and alternatives; unexplained 'best practice' or 'obvious choice' statements signal shallow analysis to a panel.
  • Answer scale-out scenario questions by naming the scaling unit at each layer (hosts, clusters, workload domains, VCF Instances, Edge clusters) and the management-plane and fabric checkpoints that must be re-validated.
  • Answer single-point-of-failure questions with the failure path, the detection mechanism (BFD/BGP, vSphere HA), the degraded-state capacity and the monitoring that proves it.
  • Compute a fully loaded cost per VM per month (CapEx amortised over asset life + OpEx, divided by VM count and 12) with the arithmetic shown and a sensitivity to VM count.
  • Handle a question you cannot answer by stating so, explaining how you would find out, and redirecting to validated design decisions — never by inventing a figure.

VCDX Program Overview (2026)#

Programme Structure and Renaming to "VMware Certified Distinguished Expert"

On 13 November 2025 VMware by Broadcom announced the next evolution of the credential as the VMware Certified Distinguished Expert (VCDX) — "the certification for private cloud experts" (VCF blog, Prashanth Shenoy). The stated purpose is to extend the credential beyond design architects to Architects, Administrators and Support professionals. DOCUMENTED: VCDX Certification Program Overview (Broadcom; undated PDF, retrieved 24 Sep 2026) and the 13 Nov 2025 blog, both read 24 Sep 2026. Earlier Academy guidance from 2025 that the requirements and defence format remained substantively unchanged is superseded as of 24 Sep 2026: the prerequisite chain below is new.

Certification Tracks
Three tracks — VCDX-VCF Architect, VCDX-VCF Administrator and VCDX-VCF Support — each anchored to the matching VCP-VCF and role-based VCAP-VCF credential.
Prerequisites (all tracks, per the Program Overview)
Hold a VCP-VCF for Architect (2V0-13.25), Administrator (2V0-17.25) or Support (2V0-15.25). Then pass three domain-based VCAP-VCF exams, one of which must be VCAP-VKS (3V0-24.25); the other two are electives from VCAP-VCF Automation (3V0-21.25), Operations (3V0-22.25), Storage (3V0-23.25) and Networking (3V0-25.25). Finally pass the role-based VCAP that aligns to the VCP held: VCAP-VCF Administrator (3V0-11.26), Architect (3V0-12.26) or Support (3V0-13.26). All of these exams are 60 items, 135 minutes, 300 scaled pass mark, USD 250 (Broadcom certification hub, 24 Sep 2026). Older Academy references to a separate data-centre design VCAP exam or to VCP data-centre or cloud equivalents are withdrawn — they do not appear in the current Overview.
Expert Evaluation
Architect track — submit a design and pass a panel evaluation. Administrator and Support tracks — complete a challenge lab and evaluation. As of 24 Sep 2026 the Overview states that expert evaluations are available exclusively for the Architect track; evaluation details for the other tracks "will be announced as they become available".
Existing VCDX Holders
Current VCDX holders who wish to carry the credential forward must pass the same three domain VCAPs (VCAP-VKS mandatory) plus the role-based VCAP; the Overview states that "no panel evaluation or challenge lab is required at this time".
Applications and Timeline
The 31 Aug 2026 VCF blog (Drew Nielsen, "Elevate Your Enterprise Value with New VCAP and VCDX Certifications") announces that VCDX-VCF Architect applications are open and invites candidates to submit now "to reserve your spot for the upcoming VCDX Panel Evaluation". No application windows, defence dates, review periods or scheduling lead times are published as of 24 Sep 2026; the older "2–4 weeks review, 30–90 days to defence" figures did not surface on any first-party page and are withdrawn. Registration and transcripts run through Certmetrics (cp.certmetrics.com/vmware); the Broadcom VCDX application-and-defence page carries no application link, handbook or dates.
Fees
Not published. No fee figure appears on the Program Overview, either blog or the Broadcom VCDX pages as of 24 Sep 2026. The legacy VCDX-DCV exam preparation guide (REV 1/2024) confirms only that an application fee and a defence fee exist. Earlier Academy figures for application, remote and in-person fees were unsourced and are withdrawn.

Defence Format and Design Submission

Defence Format
Legacy 2024 guide (VCDX-DCV, DOCUMENTED, docs.broadcom.com): the defence ran 120 minutes — a 75-minute oral defence (about fifteen minutes of presentation, then panel Q&A) followed, after a break, by a 45-minute design problem. Panel size was not specified. As of 24 Sep 2026 Broadcom has not published the format, duration, panel composition, voting rules or scoring of the new VCDX-VCF Architect panel; any timing used in Academy mock defences is a practice format, not the official one.
Design Submission (Architect Track)
Legacy 2024 guide: typical successful submissions ran between 100 and 300 pages including the application form and all diagrams; candidates were encouraged to submit conceptual, logical and physical design diagrams; defences were conducted in English; the design could be an actual project, a fictional one or a blend. Anonymising customer names, geographies and proprietary data (Customer-A, Site-B, Network-X) is Academy advice, not a published rule. Claims about a plagiarism-detection service, an "80 % original" threshold or a self-service submission portal did not surface on any current page and are not published. Whether the 2024 page guidance carries over to the VCDX-VCF Architect submission has not been stated.

Community and Benefits

Community
The Overview describes a global community of over 300 VCDX professionals.
Benefits
Perpetual vExpert eligibility, VMUG Advantage membership, a VCF personal-use licence, one free VCF exam per year and 50 % off further exams (Program Overview, September 2026).

Key Takeaways

  • For VCDX (Program Overview, read 24 Sep 2026): three tracks — Architect, Administrator, Support. Every track requires VCP-VCF + three domain VCAP-VCF exams (VCAP-VKS 3V0-24.25 mandatory, two electives from 3V0-21/22/23/25.25) + the matching role VCAP (3V0-11/12/13.26). Fees are not published.
  • For VCDX: Architect = design submission + panel evaluation (applications open since 31 Aug 2026; no windows or dates published); Administrator/Support = challenge lab and evaluation, not yet available. Legacy 2024 guide: 100–300 pages including application form and diagrams; conceptual, logical and physical diagrams encouraged; defence in English; design may be actual, fictional or a blend (anonymising the customer is Academy advice, not a published rule).
  • For VCDX: review periods and scheduling lead times are not published. Legacy defence (2024 guide) was 120 min = 75-minute oral defence (about fifteen minutes of presentation, then Q&A) + 45-minute design problem after a break; the new VCDX-VCF Architect panel format is unpublished as of 24 Sep 2026.

Design Document Structure & Best Practices#

Canonical Document Sections

There is no required page count (legacy 2024 guide: successful submissions typically ran 100–300 pages). For one set of indicative page budgets per section, use the Academy suggested outline in "The Distinguished Expert Program 2026 — Full Structure" below.

Executive Summary
:

Business drivers, customer challenges, project scope in layman's terms.

Proposed solution at high level (no technical details; 2-3 paragraph pitch).

Expected benefits (uptime improvement, cost savings, developer productivity gains).

Timeline, budget (rough order of magnitude), key milestones.

Panelists read this first; clarity and conciseness critical.

Current State Analysis
:

Existing infrastructure: servers, storage, networking, DR, staffing.

Performance metrics: CPU utilization, storage capacity, network bandwidth, availability SLAs.

Pain points: limitations, bottlenecks, compliance gaps, operational toil.

Include diagrams (network topology, VM inventory breakdown, data flow).

Clearly bound scope: "In-scope: compute, storage, networking in data center-1. Out-of-scope: WAN, security appliances (managed separately)".

Requirements & Constraints
:

Requirements
Functional (5-tier app deployment, Kubernetes workloads) and non-functional (99.95% uptime, <100ms latency, encryption at rest).
Constraints
Budget limits, hardware procurement timeline, staff skill level, regulatory compliance (PCI-DSS, HIPAA), existing vendor commitments.
Assumptions
"Assume customer willing to migrate to NSX from legacy VLAN design", "assume VCF Operations Suite available as licensed product".
Risk Factors
Technical (new technology adoption learning curve), organizational (change resistance), timeline (budget cutoff).

Format as table: each requirement with priority (P0/P1/P2), rationale, success metric.

Conceptual Design
:

High-level architecture diagram (boxes and lines, no IP addresses/hostname specifics).

Technology stack selection justified: Why VCF 9.0 over vSphere-only? Why NSX over VDS+HAProxy?

Layer breakdown: compute (vSphere), storage (vSAN), networking (NSX), management (VCF Operations), Kubernetes (VKS).

Multi-site strategy (if applicable): stretched segments, federation, DR RPO/RTO targets.

Security posture: micro-segmentation, encryption, RBAC, audit logging.

Scalability: future workload growth (2x, 5x sizing), expansion to additional sites.

Logical Design
:

Detailed component configuration: Tier-0/Tier-1 routing topology, BGP/OSPF policies, DFW rule hierarchy.

Storage design: vSAN cluster configuration (FTT, stripe width, deduplication), storage classes for Kubernetes, backup/restore strategy (Velero RPO/RTO).

Compute design: vSphere cluster sizing (CPU, memory, licensing), HA/DRS policies, resource pools, VM classes for Kubernetes.

Kubernetes design: TKG cluster topology (1 vs. 3 control planes), node pools, CNI selection (NSX vs. Antrea), service architecture.

Redundancy design: N+1 edge nodes, stretched segments for VM mobility, cross-site load balancing (GSLB).

Document in tables/matrix format: Component, Quantity, Model/Configuration, Redundancy, Justification.

Physical Design
:

Hardware bill of materials: specific models, quantities, deployment locations (rack, site).

IP addressing plan: management network, vSAN network, vMotion network, pod CIDR, service CIDR (document as IPAM spreadsheet).

VLAN/segment assignment: management VLANs, vMotion VLANs, guest VLANs, DMZ segments for ingress.

Physical wiring: uplink specifications (1GbE, 10GbE, 40GbE), redundancy (dual-homed NICs, LAG).

Power/cooling: UPS requirements, PDU distribution, thermal design (heat dissipation).

Network topology diagram: physical switches, edge routers, underlay fabric, BGP peering.

Implementation Plan
:

Phased rollout timeline: Phase 1 (foundation infrastructure), Phase 2 (workload migration), Phase 3 (optimize/harden).

Detailed runbook per phase: Pre-flight checks, deployment steps, validation criteria, rollback triggers.

Change management: communication plan, stakeholder sign-off gates, testing windows (UAT, production cutover).

Resource allocation: team size, external support (VMware PS, vendor engineering), budget burn per phase.

Risk mitigation: identified risks with contingency plans (e.g., "if vendor delivery delayed by 6 weeks, accelerate software design phase").

Validation Plan
:

Success criteria per phase: "Cluster stable after 72-hour burn-in", "all VMs migrated with zero downtime", "DFW rules enforce per security spec".

Test scenarios: failover tests (single node failure, network partition), load tests (peak workload simulation), security tests (policy violation detection).

Performance baselines: expected throughput, latency, IOPS; measurement methodology (tools: vSAN perf analyzer, Avi analytics, NSX flow exporter).

Sign-off procedure: customer acceptance criteria, customer sign-off gate prior to production handoff.

Appendices
:

Detailed configuration files: vSAN cluster config (fault domains, disk groups), vSphere cluster settings (HA thresholds, DRS affinity rules).
Network topology diagrams (Visio format exported to PDF): logically organised from conceptual to physical (L3 fabric, Tier-0/Tier-1 placement, VLAN/overlay segments, IP plan). Cite the source of any template material you reuse, per the legacy preparation guide; appendices support the design, they do not substitute for justification in the body.

Key Takeaways

  • Document every trade-off decision during design; panelists expect to see them. Phrases like "this was the obvious choice" or "best practice dictates" without justification suggest shallow analysis. Always explain why you chose A over B, what customer requirements drove the decision, and what residual risk you accepted.

RCAR Framework Mastery#

Requirement (R): What Must the Design Achieve?

Functional requirements: "Support 10,000 VMs across 3 sites", "Kubernetes workload isolation via namespaces", "Automated VM provisioning via Terraform".

Non-functional requirements: "99.99% uptime SLA", "<100ms pod-to-pod latency", "Encryption at rest for sensitive data", "Audit logging for compliance".

Business requirements: "Reduce OpEx by 30% via automation", "Enable developer self-service (reduce VM provisioning time from 2 weeks to 2 days)".

Each requirement must be measurable and traceable. Example: "Support 10,000 VMs" → cluster sizing calculation → hardware BoM.

Constraint (C): What Limits are Non-Negotiable?

Budget limit: "Capex budget $2M, Opex $500K/year".

Timeline: "Production go-live by Q4 2026".

Existing infrastructure: "Must integrate with Cisco ACI (not replace)", "Current vSphere 7.x cluster must remain operational during migration".

Regulatory: "PCI-DSS compliance required; payment systems must reside on isolated segments".

Staffing: "Local ops team has 2 FTE (full-time equivalent); external support limited to 3-month onsite engagement".

Technology: "Must standardize on VMware stack; no Kubernetes clusters outside vSphere".

Assumption (A): What are We Taking for Granted?

Organizational: "Customer is committed to VCF adoption; no political resistance to consolidation strategy".

Technical: "Existing network underlay is stable 10GbE fabric; no Layer-2 loops or MTU mismatches".

Staffing: "Customer hiring 2 additional engineers by Q3 2026; assume skills ramp-up achievable in 3 months".

Vendor: "VMware provides PoC equipment on loan for 60 days; assume delivery by 2026-06-01".

Change management: "Assume customer accepts 2-hour maintenance window per month for patching".

Document every assumption; panelists will probe: "What if this assumption proves false? What's your contingency?"

Risk (R): What Could Derail the Design?

Technical Risk
"Loss of the NSX Federation Global Manager (active and standby clusters both down). The Global Manager is management plane only — it is not required for control-plane or data-plane connectivity (VCF 9.0.1, VCF-NSX-GM-RCMD-CFG-004), so inter-site traffic and workloads keep running; what stops is any change to global segments, Tier-0/Tier-1 and firewall policy. Likelihood: low (two GM clusters, standby auto-activates when the active fails, p.2640). Impact: medium (no global policy changes until restore). Mitigation: scheduled GM backups (active GM only), documented restore-and-re-onboard procedure, quarterly restore drill. Residual risk: configuration freeze for the restore window."
Organizational Risk
"Skills gap: customer team unfamiliar with NSX micro-segmentation. Likelihood: high. Impact: security policies incorrectly applied. Mitigation: 3-week hands-on training for all ops staff + external vendor support during Phase 1. Residual risk: security misconfig by ops team in Year 2 (post-support)."
Schedule Risk
"Hardware delivery delayed 6 months due to supply chain. Likelihood: medium (current chip shortage). Impact: go-live slips to Q2 2027. Mitigation: order long-lead-time items now; negotiate expedited shipping. Contingency: phased deployment (Site-1 only in Q4 2026, Site-2/3 in Q2 2027)."
Budget Risk
"Unforeseen integration costs (legacy system API wrapping) exceeds $200K contingency. Likelihood: medium. Mitigation: detailed pre-sales discovery; allocate 15% contingency budget. Residual risk: scope reduction if costs exceed contingency."

RCAR in Defense Q&A

Panelists ask scenario-based questions to test your RCAR reasoning:

Requirement Challenge
"What if customer needs uptime to be 99.999% (five nines) instead of 99.99%? How would your design change?"

Answer: "Five nines requires sub-30 second failover and zero single points of failure. Would eliminate single edge node; require 4+ edge nodes minimum. Affects cost, complexity, and skill requirements. Would need to revisit constraint discussion (budget impact ~$400K additional)."

Constraint Violation
"Budget just got cut by 30%. What do you sacrifice?"

Answer: "Would revert to single-site deployment (eliminate federation, saves ~$400K). Would reduce node counts from 5 to 3 (loses N+2 redundancy, accept N+1). Would defer Kubernetes (VKS) to Phase 2, focusing on VM workload consolidation first. New go-live target Q2 2027."

Assumption Failure
"The hiring plan fell through; customer only has 2 FTE, not 4. Your design assumed 4 FTE for daily ops. How do you adapt?"

Answer: "Would increase automation investment (Terraform modules for VM provisioning, Aria Ops automation for capacity management). Would extend external support contract from 3 to 12 months (cost increase). Might recommend multi-cloud ops platform (Aria Ops for vSphere + cloud resources) to reduce duplicate tooling overhead."

Risk Materialization
"Both Global Manager clusters were lost in a site incident. Your design assumed a restore from backup within two hours; the ops team restored to a GM with a new IP instead of the backed-up FQDN, the Local Managers would not re-sync, and no global firewall change could be published for a day. Workloads kept running — the data plane does not depend on the GM — but a scheduled segmentation change for a go-live was missed. What went wrong?"

Answer: "Risk mitigation was adequate on paper (backup and runbook existed) but the assumption about ops-team practice was flawed: the restore prerequisites (same version, same IP or published FQDN, active-GM backup, re-onboard the standby) were never rehearsed. Should have invested in: (1) quarterly failover drills (mandatory), (2) automated failover script (no manual steps), (3) post-incident review process. Residual risk was accepted by the customer in writing after the drill results were presented — and the panel will want to see that acceptance recorded in the risk register, with the owner and the review date."

How panelists score RCAR answers: they look for (1) a requirement traced to a business driver, (2) a constraint distinguished from an assumption, (3) a risk with likelihood, impact, mitigation, residual risk and an owner, and (4) evidence that the candidate re-validated assumptions when facts changed. An answer that blames an operations team without showing the design change that removes the dependency on them scores poorly.

Key Takeaways

  • For VCDX: RCAR framework = Requirements (functional, non-functional, business) + Constraints (budget, timeline, regulatory) + Assumptions (organizational, technical, staffing) + Risks (technical, organizational, schedule, budget).
  • For VCDX: Every requirement must be measurable/traceable. Document EVERY assumption; panelists probe failure scenarios. Risk = likelihood + impact + mitigation + residual risk.
  • For VCDX: Defense Q&A tests RCAR reasoning. Scenario: "Scale to 10x?" → design adaptation (cost, complexity impact). Scenario: "Budget cut 30%?" → scope reduction options. Scenario: "Assumption fails?" → contingency plan.

The Distinguished Expert Program 2026 — Full Structure#

VCDX Renamed "VMware Certified Distinguished Expert" (13 Nov 2025)

On 13 Nov 2025 VMware by Broadcom announced the next evolution of VCDX as the VMware Certified Distinguished Expert, extending the credential to three tracks — Architect, Administrator and Support — each built on a VCP-VCF, three domain VCAP-VCF exams (VCAP-VKS mandatory) and the matching role-based VCAP-VCF. Only the Architect track (design submission and panel evaluation) is open as of 24 Sep 2026; the Administrator and Support challenge-lab evaluations are not yet available. Fees, review timelines and the new panel format are not published; see "VCDX Program Overview (2026)" above for the full prerequisite chain and sources.

Design Document Structure — Academy Suggested Outline

This is the Academy's suggested outline, not an official template: the legacy VCDX-DCV preparation guide (REV 1/2024) states there is no required minimum page or word count, lists submissions that merely echo published templates and defaults as a rejection ground, and requires any template material to be identified and cited. No design template surfaced in the 2026 Program Overview or on the Broadcom VCDX page. Section-by-section guidance (indicative sizing only):

DESIGN DOCUMENT SECTION ORDER AND INDICATIVE SIZING:

  1. Executive Summary (2-3 pages, 500-750 words)
  • Business drivers, success criteria, high-level architecture summary
  • Audience: C-level (CIO, CFO), non-technical stakeholders
  • Key statistics: workload count, cost savings, availability improvement
  1. Scope (2 pages, 400-600 words)
  • What's included in design, what's out of scope (and why)
  • Assumptions (e.g., "5-year hardware lifecycle assumed")
  • Constraints (e.g., "budget cap USD 5M", "no new hardware purchases")
  1. Assumptions, Constraints, and Risks (5-10 pages, 1500-2500 words)
  • Assumptions: each documented, impact if false (low/medium/high)
  • Constraints: customer-imposed or technical (hard stops)
  • Risks: likelihood + impact matrix, mitigation per risk
  1. Current State Assessment (10-20 pages, 2500-5000 words)
  • Existing infrastructure inventory (VMs, storage, network)
  • Utilization metrics (CPU 60-90th percentile, memory, I/O)
  • Pain points (what's broken, what's slow)
  • Growth trends (12-month capacity plan)
  1. Conceptual Architecture (5-8 pages, text + diagrams)
  • High-level business domains (Production, Development, Edge)
  • Cross-domain interconnections
  • Audience: business stakeholders, architects
  • Minimal technical detail, no IP addresses
  1. Logical Architecture (10-15 pages, text + detailed diagrams)
  • Domain-to-VCF-construct mappings (which domain = which workload domain)
  • Tier-0/Tier-1 topology, security zones (DMZ, app tier, db tier)
  • Data flow (north-south, east-west)
  • Audience: architects, security
  1. Physical Architecture (40-80 pages, very detailed)
  • Hardware (servers, storage arrays, network switches) with specific models/SKUs
  • IP ranges, VLAN/VXLAN mappings, cable diagrams
  • vSAN disk layouts, NSX node placement, edge rack location
  • Failure domain boundaries, redundancy paths
  • Detailed bill of materials (every component)
  1. Operations Design (15-25 pages)
  • Backup/restore procedure, RTO/RPO targets
  • Monitoring (tools, metrics, alerting thresholds)
  • Capacity planning (growth projections, scaling triggers)
  • Day-2 operations (runbooks, escalation paths)
  • Performance baselines, SLAs
  1. Transition/Migration Plan (8-12 pages)
  • Current state -> final state transition sequence
  • Phases (Phase 1: mgmt cluster, Phase 2: workload migration, Phase 3: decommission legacy)
  • Cutover windows, rollback procedures per phase
  1. Validation Plan (8-10 pages)
  • How to verify design meets requirements (lab testing, pilot deployment)
  • Success criteria (no more than X percent VM downtime, capacity headroom Y percent)
  • Sign-off process (customer, partners, VMware)
  1. Appendices (variable)
  • Detailed equipment datasheets, network diagrams, capacity calculations
  • RFC documents (detailed for complex decisions)
  • Bibliography/references

TOTAL ESTIMATED: 100-300 pages depending on design scope (legacy 2024 guide: typical successful submissions 100–300 pages including the application form and all diagrams)

Design Decision Masterclass

Every design decision must be documented with complete justification. Template:
Decision template (DD-XXX) — fields: Decision, Justification, Impact, Risk and mitigation, Alternatives considered, References.

DD-001: Management cluster size (4 hosts, high-availability deployment, vs 3 hosts, simple deployment)

FieldContent
DecisionDeploy a 4-host management cluster — the VCF 9.0 high-availability deployment minimum; a simple deployment allows 3 hosts with vSAN (VCF 9.0.1 p.1331)
JustificationCustomer expects no reduction in protection during host maintenance. With 4 hosts, one host in maintenance mode still leaves FTT=1 compliance and vSphere HA failover capacity for the management VMs; a 3-host cluster runs degraded (no rebuild target, HA admission control at risk) for the whole update window (customer risk tolerance low).
ImpactAbout 33% more management hosts than the 3-host simple model (4 vs 3); one host of compute is held for N+1
Risk and mitigationCost of the 4th host. Mitigation: show that 4 hosts is the documented high-availability minimum (VCF 9.0.1 docs p.1331); offer the 3-host simple model only where the customer accepts its reduced availability during maintenance.
Alternatives considered3-host simple deployment (cost-optimal; degraded protection during each host update unacceptable to customer); 5-node HA (excessive, diminishing returns on availability)
ReferencesVMware HA best practices, customer RTO requirement (4h max), cost analysis

DD-002: vSAN capacity tier (SAS vs NVMe for all-flash)

FieldContent
Decision10TB SAS capacity tier (instead of all-NVMe like 3.8TB NVMe)
JustificationWorkload mix: 70% cold data (backup archives, infrequent access), 30% hot data. All-NVMe = cost per TB 5x higher, wasted on cold data. 10TB SAS + NVMe cache tier = 80% cost savings, latency acceptable for cold-data workloads.
ImpactCost USD 500K saved, slightly higher latency for cold data (10ms vs 1ms)
Risk and mitigationUnderestimation of hot-data percentage (if real 50% hot, SAS becomes bottleneck). Mitigation: re-baseline workload in Y2, upgrade to NVMe if needed.

Additional Hands-On Practice Labs (VCF 5.2 / 9.x)

Targeted practice labs for deeper mastery of Vcdx topics on VCF 5.2 and VCF 9.x.
🔬 Lab C1: Build a VCDX-Grade Design Document for VCF 9.0 Brownfield Migration

VCF Version:

VCF 5.2 → 9.0

Objective:

Produce a 120–180 page VCDX-style design document for a real-world VCF 5.2 → 9.0 migration.

Write the Executive Summary with scope, goals, non-goals (2–3 pages).

Fill out a full RCAR register (Requirements, Constraints, Assumptions, Risks): minimum 25 R, 10 C, 15 A, 10 Risks.

Create conceptual, logical, and physical designs for compute, storage, network, management (sized per the indicative outline above).

Add a Design Quality matrix: Availability, Manageability, Performance, Recoverability, Security — rate each design decision.

Write explicit justification for every design decision with references (≥5 references per major area).

Include operational procedures (runbooks) as an appendix.

Add risk register with mitigation + acceptance owner per risk.

Validation:

Document passes self-review against the Academy self-review checklist (not an official rubric — VMware does not publish how the VCDX defence is scored); every DD has RCA (Rationale, Consequence, Alternatives).
🔬 Lab C2: Mock VCDX Defense — Timed Session (Academy practice format)

VCF Version:

VCF 9.0

Objective:

Simulate a VCDX Distinguished Expert defence against a panel using the Academy practice format, not the official one.

Recruit 2 peers as panelists; share your design doc 24h before.

Run a 75-minute timed mock defence: fifteen minutes of presentation, 45 minutes of panel Q&A, fifteen minutes of design scenario (Academy practice format). For reference, the legacy 2024 guide describes a 120-minute defence — a 75-minute oral defence followed, after a break, by a 45-minute design problem; the new VCDX-VCF Architect panel format is unpublished as of 24 Sep 2026.

Record video for self-review.

Answer at least 20 design-probing questions with rationale and alternatives.

Score yourself against an Academy self-review checklist (not an official rubric — VMware does not publish how the VCDX defence is scored): requirement-to-decision traceability, justification of alternatives, risk handling, clarity.

Iterate: redo weak areas and run a second defense within 72 hours.

Validation:

You can defend every decision with R/C/A context; no panic stalls >10s; panelists can't find holes after 30min probing.
🔬 Lab C3: Design Scenario — Regulated Industry (Healthcare + Middle East Data Residency)

VCF Version:

VCF 9.0

Objective:

Design a VCF 9.0 architecture for a UAE healthcare customer subject to UAE health-data regulation: federally, Federal Law No. (2) of 2019 on the use of ICT in health fields (the Ministry of Health and Prevention, MOHAP, is the law's 'Ministry'); in Abu Dhabi, the Department of Health (DoH) ADHICS standard, now in version 2.

Capture constraints: UAE health-data residency (Federal Law No. (2) of 2019, Article 13: health data and information related to health services provided in the UAE may not be stored, processed, generated or transferred outside the State unless a Health Authority resolution, issued in coordination with MOHAP, allows it), retention of health data for at least 25 years from the last health procedure (Article 20), ADHICS V2 compliance if the customer operates in the Emirate of Abu Dhabi, and any cybersecurity standard set by the customer's own health authority.

Design isolated management and workload domains with NSX + vDefend zero-trust.

Size vSAN ESA for 700 workloads with 3-year growth; pick OSA vs ESA vs Max with rationale.

Design HCX RAV migration waves from source vSphere 7.x and write cut-over runbook.

Design site availability and DR as two separate decisions: (a) a vSAN stretched cluster for site availability within one metro region (less than 5 ms RTT between the data sites, VCF 9.0.1 docs p.1609), and (b) DR to a secondary site inside the UAE with a stated RPO/RTO and a replication-based recovery design; add NSX Federation where networks must span the locations.

Document which compliance controls each design decision satisfies.

Validation:

Design satisfies every listed compliance control with explicit trace; sizing is defensible with math.
🔬 Lab C4: Design Scenario — Sovereign Cloud Provider (Saudi Arabia PIF Style)

VCF Version:

VCF 9.0

Objective:

Design a multi-tenant VCF 9.0 sovereign cloud with HCX-based onboarding.

Design Instance/Fleet topology for 5 customer tenants with hard isolation.

Choose tenancy model: Private Cloud per tenant vs shared infra with DFW isolation.

Design HCX service mesh onboarding per tenant with isolated network extension, and document the isolation boundary each tenant depends on.

Validation:

Every tenant boundary is traceable to explicit compute, storage and network design decisions with rationale.

Key Takeaways

  • Scenario Q: "Scale to 10x"— Panelist: "Your design supports 5000 VMs. What if customer grows to 50,000 VMs?" Model Answer: "Physical cluster scaling: add hosts (vSAN adds capacity linearly). Management cluster: add RM cluster (federation scales horizontally). Network: edge scale adds multiple T0 instances and Edge clusters per workload domain. Beyond one VCF Instance's limits, add VCF Instances to the fleet and federate NSX (Global Manager) only where cross-site segments are required. Re-validate the management plane (VCF Operations node size, SDDC Manager limits) and the physical fabric (leaf-spine ports, TEP subnets) at each step — scaling is a redesign checkpoint, not a hardware order."
  • Risk Q: "Single Point of Failure"— Panelist: "What if the primary data center WAN link fails?" Model Answer: "Design includes 2 ISP uplinks (primary + backup, active-active load-share via BGP). If primary fails, BGP withdraws routes, backup carries 100% traffic (slight congestion, but service continues). BFD on the eBGP sessions detects the failure in under a second; the Tier-0 keeps its ECMP paths via the second ToR pair. Quantify the degraded state (bandwidth, affected SLAs) and name the monitoring that alerts on it."
  • Cost Q: "Cost per Workload"— Panelist: "What's the fully-loaded cost per VM/month?" Model Answer: "CapEx: USD 10M hardware (3-year life) = USD 3.33M/year. OpEx: USD 1M/year (staffing, support, power). Total: USD 4.33M/year. VM count: 5000 average. Cost per VM/month: USD 4.33M / 5000 / 12 ≈ USD 72/VM/month — then show the sensitivity: at 4,000 VMs it is USD 90, at 6,000 VMs USD 60, so utilisation, not hardware price, drives the unit cost."
  • "I don't know" Tactical Response— If panelist asks question you can't answer: (1) Don't guess, admit "I don't know off the top of my head, but here's how I'd investigate." (2) Explain methodology (e.g., "I'd check NSX performance guide, measure in lab, consult field engineers"). (3) Redirect to known ground — relate the question to a design decision you did validate. Never invent a number; panels probe invented figures until they collapse.

Exam Mapping: VCDX-VCF Architect — VMware Certified Distinguished Expert — VCF Architect

  • Design enterprise-scale VCF architecture
  • Defend design decisions with RCAR methodology
  • Demonstrate operational awareness for day-2 management
  • Address availability, security, and scalability requirements

Labs in This Section

References

Was this page useful?
Type to search. ↑ ↓ to move, Enter to open, Esc to close.