VCF Operations Dashboard Customization & Capacity Planning
Objectives
- Navigate VCF Operations console and understand its role in day-2 operations
- Design and implement custom dashboards that align with VCDX design intent (capacity, compliance, availability)
- Configure threshold-based alerting and automated remediation triggers for VCF workload domains
- Analyze actual utilization vs. design projections to validate cluster sizing assumptions
- Use VCF Operations reporting to build VCDX design artifacts (capacity headroom proof, compliance posture)
Prerequisites
VCF 9.0 lab environment fully deployed with management domain + 1 workload domain (4 hosts minimum). VCF Operations console must be accessible (https://vcf-operations-ip (TCP 443)). At least 7 days of baseline metrics available (or manually seeded test data). DNS resolution configured for vcf.local domain.
Prior labs: holodeck-02 or equivalent VCF deployment
Required skills:
- VCF architecture fundamentals (management domain, workload domains, SDDC Manager)
- Basic understanding of capacity planning concepts (CPU oversubscription ratios, memory overhead, storage utilization)
- Comfort with vSphere Client and VCF management console UIs
- Ability to interpret performance metrics and troubleshoot resource bottlenecks
Lab Environment
VCF Management Domain (SDDC Manager, vCenter, NSX Manager) + Workload Domain A (4 ESX hosts, shared vSAN datastore). VCF Operations is deployed in the Simple model (single primary node) in the management domain, accessible via FQDN ops.vcf.local or direct IP. Network: 10.0.0.0/20 for management services, 10.1.1.0/24 for ESX hosts.
Credentials
| System | Username | Password |
|---|---|---|
| SDDC Manager | administrator@vsphere.local | VCF admin password from deployment spec |
| vCenter Server | administrator@vsphere.local | Same as SDDC Manager |
| VCF Operations | admin | VCF Operations admin password from deployment spec (default: same as SDDC Manager) |
| NSX Manager | admin | NSX admin password from deployment spec |
Tasks
Task 1 Access and Navigate VCF Operations Console
ManageabilityEstablish familiarity with VCF Operations role in the VCF operational lifecycle. VCF Operations is the single-pane-of-glass for capacity, compliance, and performance visibility — critical for any VCF design that must demonstrate 'what gets monitored' and 'how it gets managed'. A VCDX panelist will ask: 'Walk me through your day-2 operational dashboards. What metrics are you tracking and why?'
Open a web browser and navigate to https://ops.vcf.local (or https://<vcf-operations-ip> if DNS is not configured). Accept the self-signed certificate.
Log in with username 'admin' and the VCF Operations password (use the same password as SDDC Manager administrator@vsphere.local).
Navigate to Administration > Integrations and open the Accounts tab. Verify that the vCenter, NSX and SDDC Manager accounts for the VCF instance show a healthy collection status.
Navigate to Inventory. Expand the tree to see: Workload Domains > [Workload Domain A] > Clusters > Cluster A > Hosts.
Navigate to Infrastructure Operations > Dashboards & Reports and click Dashboards > All. Review the predefined dashboard categories (for example Availability, Configuration, Inventory, Performance, Capacity), then expand VMware Infrastructure Health and open VMware Infra Management Health Overview.
In VMware Infra Management Health Overview, inspect the widgets: VMware Infrastructure Health reports availability, services, virtual machines, certificates, passwords and active alerts for each configured VCF account and its domains. Note the time range selector (top right) — ensure it's set to 'Last 7 Days' or 'Last 24 Hours' depending on available data.
Validation Gate
Check: VCF Operations console is accessible and showing inventory + at least one predefined dashboard with data
Expected: You can navigate to Inventory, see 1 workload domain / 1 cluster / 4 hosts, and view the VMware Infra Management Health Overview dashboard with data populated
Common Errors
Task 2 Design a Custom Dashboard for Capacity Planning & Design Validation
ManageabilityIn VCDX, you must articulate 'how will I know if my design is working as intended?' This task forces you to identify 3-4 key capacity and availability metrics that directly map to your design assumptions. For example, if you designed the cluster for 4:1 CPU oversubscription, you must monitor actual CPU contention (Ready % or %Ready) and prove it stays below design limits. This is the difference between a theoretical design and a validated deployment.
Navigate to Dashboards > Create Dashboard. Enter the dashboard name: 'VCF Design Capacity Baseline' and description: 'Validates sizing assumptions: CPU oversubscription, memory headroom, storage utilization, and DRS/HA availability.'
Add Widget 1 (CPU Utilization): Click 'Add Widget' > Select 'Line Chart' > Configure as follows:
- Data Source: vCenter (management domain) > Cluster A
- Metrics: CPU Utilization (%), CPU Ready (%)
- Time Range: Last 7 Days
- Y-axis label: 'CPU %'
- Title: 'CPU Utilization & Ready (7-day baseline)'
Thresholds: Green (0-60%), Yellow (60-75%), Red (>75%)
Add a horizontal reference line at 75% (your design's CPU threshold).
Add Widget 2 (Memory Utilization): Click 'Add Widget' > Select 'Line Chart' > Configure as follows:
- Metrics: Memory Utilization (%), Memory Swapped (%)
- Time Range: Last 7 Days
- Thresholds: Green (0-70%), Yellow (70-85%), Red (>85%)
- Title: 'Memory Utilization & Swap Activity (7-day baseline)'
If Memory Swapped > 0%, it indicates memory pressure and possible VM performance impact.
Add Widget 3 (vSAN Capacity):
- Type: Pie Chart
- Data Source: vCenter > Cluster A > vSAN Datastore
- Metrics: Used Capacity (GB), Free Capacity (GB)
- Add a data label showing percentages
- Title: 'vSAN Storage Utilization (Current Snapshot)'
If used > 80%, change pie slice color to yellow; >90%, red.
Add Widget 4 (DRS & HA Status):
- Type: 'Multi-stat Gauge' or 'Scorecard'
- Metrics: (a) DRS Recommendation Count (should be ≤2 per day for balanced cluster), (b) HA Protected VMs (%), (c) Host Failures Tolerated (should be ≥1)
- Title: 'DRS & HA Availability Status'
- Thresholds: HA Protected <95% = Yellow, DRS recommendations >10 = Yellow.
Save the dashboard. Click 'Save' and confirm the name 'VCF Design Capacity Baseline'.
Validation Gate
Check: Dashboard 'VCF Design Capacity Baseline' exists with 4 widgets (CPU, Memory, vSAN, DRS/HA), all displaying data
Expected: All widgets show at least 1 data point from the last 7 days; thresholds are color-coded; you can articulate what each metric means for cluster health
Common Errors
Task 3 Configure Compliance & Alerting Policies for VCF Design Governance
ManageabilityVCDX requires you to define and monitor compliance with your design. This task operationalizes compliance: you identify the design rules (e.g., 'CPU Ready must stay <5%', 'vSAN capacity must not exceed 80%', 'DRS must maintain load balance'), configure automated alerts, and demonstrate how you'll be notified if the design assumptions are violated. This is the 'feedback loop' that keeps production designs healthy.
In VCF Operations, navigate to Infrastructure Operations > Configurations and click the Alert Definitions tile. Click Add to create an alert definition (base object type, impact, criticality), then add symptom definitions and recommendations and assign the definition to your policy.
Create Alert Definition 1: CPU Oversubscription Risk
- Alert Name: 'Cluster CPU Ready Threshold Exceeded'
- Condition: Cluster A > CPU Ready (%) > 5 (for more than 10 minutes)
- Severity: Warning (Yellow)
- Notification: email to 'your-email@company.com' via a notification rule (Task 2 step 6)
- Description: 'Triggers when CPU ready exceeds 5%, indicating compute contention. Design assumed 4:1 CPU oversubscription with <5% ready as acceptable threshold.'
Create Alert Definition 2: Memory Pressure Detection
- Alert Name: 'Cluster Memory Swapped Detected'
- Condition: Cluster A > Memory Swapped (%) > 0.1 (for more than 5 minutes)
- Severity: Critical (Red)
- Notification: email plus a Slack or ServiceNow outbound instance via a notification rule
- Description: 'Memory swap should never occur in a properly sized VCF cluster. This alert indicates either a memory leak, VM memory balloon driver failure, or cluster undersizing.'
Create Alert Definition 3: vSAN Capacity Threshold
- Alert Name: 'vSAN Storage Capacity Warning'
- Condition: Cluster A > vSAN Capacity > 80%
- Severity: Warning (Yellow) for >80%, Critical (Red) for >90%
- Recommendation (text): 'Provision additional vSAN capacity or implement storage tiering'; notification by email
- Description: 'vSAN clusters should maintain <80% capacity to allow for data growth and rebuilds. At >90%, cluster is in critical capacity state and one host failure may cause data loss.'
Create Alert Definition 4: DRS Load Balance Deviation
- Alert Name: 'Cluster Load Imbalance Detected'
- Condition: Cluster A > DRS Recommendation Count > 10 (per day)
- Severity: Warning (Yellow)
- Recommendation (text), notification by email: 'Review VM placement. Possible causes: (1) unbalanced VM sizing, (2) affinity rules preventing DRS, (3) maintenance mode on a host.'
- Description: 'More than 10 DRS recommendations per day indicates the cluster is not staying balanced. This may impact performance and HA redundancy.'
Navigate to Infrastructure Operations > Configurations and click the Outbound Settings tile. Add a Standard Email plug-in instance (SMTP host, port, sender, authentication) and confirm it is running. Then click the Notifications tile, click Add, and create a notification rule: Notification Type = Alert; criteria = your four alert definitions on Cluster A; outbound method = the Standard Email instance with your recipient.
Validation Gate
Check: Four alert definitions are active in the policy: CPU Ready, Memory Swap, vSAN Capacity, DRS Load Balance. A Standard Email outbound instance and a notification rule are configured.
Expected: Infrastructure Operations > Configurations > Alert Definitions lists the four definitions; Outbound Settings shows the Standard Email instance running; Notifications shows your rule as active.
Common Errors
Task 4 Create a Compliance Report: Design vs. Reality
ManageabilityThis is the VCDX artifact: a formal report that proves your design is working as intended. You'll compare your design assumptions (capacity projections, availability targets, cost models) against actual metrics collected over 7 days. This is the 'proof in the pudding' — design is only valid if it holds up under real workload. A panelist will ask: 'Show me your KPIs. Is the cluster performing within your design envelope?'
Create a new dashboard called 'Compliance Report: Design Validation' (or export the 'VCF Design Capacity Baseline' dashboard as a starting point). This dashboard will serve as the basis for your compliance report.
Add a title widget (text box) at the top of the dashboard with the following text:
'COMPLIANCE REPORT: VCF 9.0 Design Capacity Validation
Design Period: [Enter current week]
Cluster: Workload Domain A, Cluster A
Design Intent: 4:1 CPU oversubscription, 70% memory utilization target, 80% storage capacity target, DRS load balance, HA coverage >95%'
Add a fourth widget: 'Design vs. Actual Scorecard'. Create a multi-stat widget with the following KPIs:
- 'CPU Utilization (7-day avg)': Design projection was X%, Actual is Y%. Pass/Fail: ✓ if Y ≤ X + 10% headroom
- 'Memory Utilization (7-day avg)': Design projection was X%, Actual is Y%. Pass/Fail: ✓ if Y ≤ 70% and Swap = 0%
- 'vSAN Capacity': Design projection was X%, Actual is Y%. Pass/Fail: ✓ if Y ≤ 80%
- 'HA Coverage': Design requirement was >95%, Actual is Y%. Pass/Fail: ✓ if Y ≥ 95%
Export the dashboard as PDF: In the dashboard view, click the menu (three dots) > Export as PDF. Name the export 'VCF-Design-Compliance-Report-<date>.pdf'.
Review the exported PDF and write a brief analysis (in your lab notes or a separate document):
- Are actual metrics within your design envelope?
- If not, identify the variance and root cause (e.g., 'CPU utilization 15% higher than projected due to unexpected batch job from Finance team').
- What design changes would you make if re-running this exercise with the actual utilization data?
- Did any alerts fire during the 7-day period? If yes, document which alerts and what triggered them.
Archive the dashboard and PDF export in your VCDX design artifacts folder. Add a README.md file with the context: 'This report validates the VCF 9.0 cluster design (capacity, availability, performance) against 7 days of production metrics.'
Validation Gate
Check: Compliance report dashboard exists with title, 4 KPI widgets, and scorecard. PDF export is available. Written analysis (3-5 paragraphs) documents design vs. actual variance.
Expected: You have a formal compliance report artifact that proves your design is working as intended, suitable for VCDX design defense.
Common Errors
Task 5 Set Up Automated Actions, Notifications and Escalation
ManageabilityA mature VCF design includes automated responses to known issues and a clear hand-off to humans. In VCF Operations this is built from three documented mechanisms: actionable recommendations on alert definitions that a policy can automate, notification rules that route alerts to outbound plug-ins (email, Webhook, Slack, ServiceNow, SNMP), and Automation Central jobs for scheduled reclaim and rightsizing. Escalation timing (who is paged after 15 or 30 minutes) lives in the incident tool that receives the notification, not in VCF Operations. This task moves you from 'we monitor it' to 'the system acts where it safely can and notifies the right people otherwise'.
Navigate to Infrastructure Operations > Configurations and click the Recommendations tile. Review the built-in recommendations and note which include an Action (for example power off or resize actions on VMs). Click Add to create a text recommendation for your 'Cluster Load Imbalance Detected' alert: 'Review VM placement: check affinity rules, host maintenance mode and VM sizing; run DRS recommendations from the vSphere Client if the cluster is manual or partially automated.'
Attach recommendations to your four alert definitions: open Infrastructure Operations > Configurations > Alert Definitions, edit each definition and add the relevant recommendation (text or action). For the vSAN capacity alert add a text recommendation: 'Add vSAN hosts or capacity, review snapshot and powered-off VM reclaim, and open a capacity planning change.'
Open Infrastructure Operations > Configurations > Policies, edit the policy applied to Cluster A and open its Alerts and Symptoms settings. For an alert definition whose first-priority recommendation has an action (for example a VM rightsizing alert), set Automate to Local so the action runs when the alert triggers. Leave your four custom alerts non-automated.
Create the escalation path with outbound instances and notification rules. In Infrastructure Operations > Configurations > Outbound Settings add a second instance (Webhook Notification Plugin pointing at your incident tool, or a Slack or Service-Now plug-in instance). Then in the Notifications tile add a rule: Notification Type = Alert; Criticality = Critical; Notify On = New; outbound method = the Webhook/Slack/ServiceNow instance. Keep the email rule from Task 3 for Warning alerts.
Schedule safe reclaim automation: click Operations > Automation Central and create a job that deletes snapshots older than 14 days on Cluster A (Reclaim > Delete old snapshots), runs weekly, and emails a report. Review the Schedule calendar to confirm the job.
Validation Gate
Check: Four alert definitions carry recommendations; one policy entry has Automate = Local; two notification rules route Warning alerts to email and Critical alerts to the incident tool; one Automation Central job is scheduled.
Expected: Alerts raise with recommendations attached, Critical alerts reach the external incident tool for timed escalation, and snapshot reclaim runs on schedule. Source: VCF 9.0.1 docs pp.3157-3158, 3193-3199, 3205, 3349-3351.
Common Errors
Final Validation
VCF Operations is now configured as a complete operational platform: dashboards validate design assumptions, alert definitions and policies enforce SLAs, recommendations and policy automation act where safe, notification rules hand critical alerts to the incident tool, and Automation Central reclaims capacity on schedule.
✓ VCF Operations console is accessible and showing 4-host workload domain with healthy metrics → Inventory shows 1 workload domain, 1 cluster, 4 hosts; Data sources all report OK
✓ Custom dashboard 'VCF Design Capacity Baseline' exists with 4 widgets: CPU, Memory, vSAN, DRS/HA → All widgets show 7-day trend data; thresholds are color-coded
✓ Four alert definitions are active for Cluster A: CPU Ready, Memory Swap, vSAN Capacity, DRS Load Balance → Definitions listed under Infrastructure Operations > Configurations > Alert Definitions and activated in the Cluster A policy
✓ Compliance report dashboard and PDF export exist with design vs. actual scorecard → PDF file is available with title, metrics, and design validation scorecard
✓ Recommendations attached to the four alert definitions; one policy entry automated (Automate = Local); one Automation Central reclaim job scheduled → Alert Definitions show recommendations; the policy shows the automated definition; Operations > Automation Central shows the job on the calendar
✓ Notification rules route Warning alerts to email and Critical alerts to the incident tool (Webhook, Slack or ServiceNow instance) → Infrastructure Operations > Configurations > Notifications lists both rules as active; Outbound Settings shows both instances running
Cleanup / Restore
Snapshot: vcp-admin-01-complete
• Disable test alert policies if any were created for testing
• Take a snapshot of VCF Operations configuration state for future labs: 'vcp-admin-01-complete'
• Archive dashboard exports and compliance reports in your VCDX design artifact folder
• Document any design insights discovered during compliance validation (e.g., 'CPU utilization was 15% higher than projected due to [reason]')
Design Reflection (VCDX)
VCDX panelists probe your operational readiness: 'How do you know your design is working? Show me your dashboards, your alerts, your playbooks.' This lab answers that question with artifacts: a compliance report proving metrics align with design intent, alert policies enforcing SLAs, and remediation workflows demonstrating architectural maturity. A panelist examining this lab will ask: (1) Why those specific metrics? (Tie to design assumptions: e.g., CPU Ready because you designed 4:1 oversubscription) (2) What's your alert tuning strategy? (Aggressive alerts = false positives and alert fatigue; soft thresholds = latent failures. Justify your choices.) (3) How do you prevent alert fatigue? (Escalation policies, deduplication, context-aware thresholds.) Be prepared to defend your threshold values — '5% CPU Ready' — by referencing VMware best practices and your own capacity modeling.
Requirements
- VCF Operations console must provide single-pane-of-glass visibility into cluster capacity, compliance, and performance
- Operational dashboards must directly reflect design assumptions (oversubscription ratios, utilization targets, availability requirements)
- Alerts must be configured to detect and report design violations (e.g., CPU contention, capacity threshold breach)
- Alert routing must follow a tiered model: Warning alerts to the operations mailbox, Critical alerts to the incident tool, with timed escalation to on-call handled by that tool
- Automation must be limited to safe, reversible actions (policy-automated recommendations, scheduled reclaim); destructive actions require manual confirmation
Constraints
- VCF Operations deployed in the Simple model is a single primary node with no application-level HA (vSphere HA restarts it); the High Availability model (primary, replica and data nodes) or Continuous Availability must be chosen at deployment if monitoring must survive a node failure (VCF 9.0 Design, Table 15)
- Metrics collection has a 5-minute default interval; alerts cannot detect sub-5-minute events
- Notification channels (email, SMS, PagerDuty) depend on external integrations and network connectivity
- Dashboard export to PDF captures a static snapshot; real-time trends require the dashboard UI
- VCF Operations automates only actions attached as first-priority recommendations and enabled in the policy; other remediation happens in the external system that receives the notification (Webhook, Slack, ServiceNow, REST)
Assumptions
- Operations team has 24/7 on-call coverage and can respond to paged alerts within 30 minutes
- Email, Slack, and PagerDuty are integrated and configured before alerts are deployed to production
- Design assumptions (CPU oversubscription ratio, memory utilization target, capacity thresholds) are documented and agreed upon before deployment
- 7-day baseline period is sufficient to establish normal operating patterns and tuning alert thresholds
- Workload on the cluster is stable during the compliance reporting period (no major new deployments or business changes)
Risks
- Alert threshold misconfiguration causes either alert fatigue (too aggressive) or missed failures (too lenient) — IMPACT: operational efficiency loss or late detection of design violations — MITIGATION: start with soft thresholds, monitor alert volume for 1 week, then tune to optimal false-positive rate (<5 per day)
- VCF Operations outage (Simple model) prevents all operational visibility — IMPACT: blind operations until the node restarts — MITIGATION: deploy the High Availability model for production, export dashboards regularly as PDF, keep SNMP/syslog forwarding from vCenter as backup monitoring
- Notification to the incident tool fails silently (outbound instance stopped, webhook authentication failure) — IMPACT: alert condition persists, no escalation occurs — MITIGATION: test outbound instances monthly, enable the notification heartbeat for critical rules, audit the incident tool for missing alerts
- Threshold tuning is never completed — operations runs with default thresholds that don't match workload characteristics — IMPACT: irrelevant alerts, missed real issues — MITIGATION: schedule monthly threshold review; adjust based on actual utilization trends
Self-Assessment Discussion Prompts
- Your design assumes 4:1 CPU oversubscription with <5% CPU Ready as acceptable. Walk me through how you arrived at these numbers. What workloads are running? What's the business impact if CPU Ready hits 10%?
- You configured an alert for CPU Ready >5%. How often do you expect this alert to fire in steady state? If it fires daily, is your threshold too aggressive?
- VCF Operations default metric collection interval is 5 minutes. This means a 2-minute CPU spike might be missed. How does this affect your alerting strategy? Should you increase collection frequency?
- Your policy automates a rightsizing action when an alert fires. What happens if the action fails or is not applicable (for example the VM has a reservation)? How does your notification path cover that failure case?
- You're comparing 7-day actual metrics against design projections. Your actual CPU utilization is 72%, but you projected 60%. Is this a design failure or a measurement artifact? How would you investigate?
- If vSAN capacity exceeds 80%, your notification rule opens a ticket in the incident tool. Capacity planning can take 4-8 weeks. What do you do in the meantime — Automation Central reclaim, storage policy changes, or telling users you are out of space?
- Memory swap detected (critical alert). Your remediation is to send an email. But by the time the email is read, 100 VMs might already be swapping. Is 'send email' sufficient, or do you need a more aggressive auto-remediation?
Extensions
Advanced Dashboard: Workload Domain Performance Baseline
Create a second dashboard focused on workload domain performance: VM density per host (VMs/core), application response time (if APM integrated), storage latency (read/write milliseconds), network throughput (vSAN replication, NSX TEP). This dashboard proves that your workload domain can sustain the designed VM density without degradation.
moderateVCF 5.2.x Operational Comparison
If you have access to a VCF 5.2.x environment, document the differences between VMware Aria Operations 8.x (VCF 5.2) and VCF Operations 9.0: different UI (Visualize/Operations/Infrastructure Operations menus), metric names, policy and alert workflow. Compare the dashboard layout and alerting capabilities between VCF 5.2.x and 9.0. Document 5 specific operational differences.
sameIntegrate Custom Metrics from VCF Automation
If VCF Automation is deployed in your lab, configure VCF Operations to collect VCF Automation metrics (e.g., provisioning success rate, cost per VM, tenant quota utilization). Add a dashboard widget showing VCF Automation health and tenancy compliance. This extends VCF Operations visibility beyond infrastructure into application/tenancy operations.
harderCost Allocation & Chargeback Dashboard
Create a dashboard that tracks cost per workload domain and per tenant (if multi-tenant environment). Metrics: CPU hours consumed × $/vCPU-hour, Storage GB × $/GB-month, Network egress × $/GB. Demonstrate how VCF Operations enables cost-based capacity planning decisions. This is highly relevant for VCDX scenarios involving cost optimization.
harderReferences
- VMware Cloud Foundation 9.0 Operations and Administration GuideTier 1 — Official
Official Broadcom documentation for VCF operations, the VCF Operations console, dashboard creation, alert definitions, notifications, outbound settings and Automation Central. - VCF 9.0 Planning and Preparation WorkbookTier 1 — Official
Capacity planning reference: CPU oversubscription, memory overhead and storage utilisation inputs. The 4:1 vCPU:pCPU and 5% CPU Ready figures used in this lab are example design assumptions to validate against your workload, not values taken from this workbook. - VCF 9.0 documentation — vSAN 9.0 storage design and capacity planningTier 1 — Official
Deep dive on vSAN capacity model: FTT overhead, rebuild capacity, compression, deduplication impact on usable capacity. Critical for understanding your vSAN 80% threshold. - vSphere DRS and HA documentation (VCF 9.0)Tier 1 — Official
vSphere DRS and HA documentation for DRS migration threshold levels and HA admission control. The 10-recommendations-per-day alert threshold used in this lab is an example value to tune, not a VMware recommendation. - William Lam — VCF Operational ExcellenceTier 3 — Background reading
Expert blog on VCF day-2 operations, monitoring strategies, performance tuning. Often covers VCF Operations configuration tips and gotchas.