If you’re evaluating a VMware alternative, monitoring is one of the first topics that comes up. It’s a capability that quietly blocks a migration if it’s missing. Your team already has dashboards, alerting rules, and capacity planning workflows built around vCenter and vROps. The question isn’t whether Private Cloud Director (PCD) has monitoring. It’s whether the monitoring it provides fits into how your team actually operates.
This post covers how PCD’s built-in monitoring stack works, how it compares to the vSphere monitoring capabilities your team already relies on, and how to integrate PCD metrics into the observability platform you’re already running.
Background
VMware administrators typically rely on vCenter performance charts for real-time host and VM metrics, vRealize Operations (vROps, now Aria Operations) for capacity planning and alerting, and ESXTOP for deep per-host diagnostics. These tools are tightly integrated with the vSphere stack but are also proprietary and licensed separately.
PCD takes a different approach. Rather than a proprietary monitoring suite, PCD exposes metrics through open, standards-based endpoints that work with any VictoriaMetrics/Prometheus-compatible scraper. Built-in Grafana dashboards provide immediate visibility, and the exporter architecture lets your team integrate PCD metrics into whatever observability platform you already run.
How the Monitoring Stack Works
Monitoring Architecture
PCD’s monitoring stack is provisioned automatically per region, with no additional installation or configuration required. In vSphere terms, think of this as if vCenter’s performance monitoring came pre-configured on every host the moment you add it to the cluster, with no separate vROps deployment or license.
The stack consists of VictoriaMetrics-compatible exporters running as system services on every hypervisor host, and a Grafana instance providing pre-built dashboards accessible from the PCD UI. Metrics collection begins automatically when the hypervisor role is assigned to a host.
How Metrics Are Collected
Each hypervisor host runs two exporters: node-exporter for host-level metrics (CPU, memory, disk, network) and libvirt-exporter for VM-level metrics. If you’re used to the data you get from ESXTOP or vCenter’s per-host performance charts, the raw exporter endpoints on each host provide a comparable level of detail.
A VictoriaMetrics agent on each host gathers metrics from both exporters locally, applies recording rules, and sends processed output to a VictoriaMetrics instance on the PCD management plane every 5 minutes. Labels identifying the originating host UUID and VM UUID are added during this remote write.
An important distinction: the raw exporters on each host expose the complete set of metrics. The VictoriaMetrics agent applies recording rules before forwarding, so the management plane holds a reduced, aggregated set. Organizations that require full metric fidelity should scrape the per-host endpoints directly (see Exporting PCD Metrics below).
Available Metrics
PCD collects metrics at both the hypervisor level and the VM level. Hypervisor metrics cover compute capacity and usage, storage throughput per device, and network throughput for the host. VM metrics go deeper: per-vCPU throttling, memory allocation and usage, per-disk IOPS, throughput and latency, and per-interface packet drop rates. This maps closely to what vCenter exposes in its performance charts and what ESXTOP provides in real-time, with the difference that PCD metrics are available through standard exporters rather than a proprietary interface.
The full list of metric names, namespaces, and labels is available in the PCD Metrics Reference.
Accessing Grafana
Grafana is accessible directly from the PCD UI, providing per-region views of all hypervisor and VM metrics. Custom dashboards can be created within Grafana using the standard dashboard editor. This is the closest equivalent to vCenter’s built-in performance charts, with the added flexibility that Grafana’s query language and visualization options are more extensible than vCenter’s fixed chart views.
For navigation details, see the PCD Monitoring documentation.
Service Health Dashboard
Separate from the Grafana-based metrics, PCD includes a Service Health Dashboard in the UI. This serves a similar function to the vSphere Health Checks, providing a quick operational status view without requiring you to dig into metrics. The dashboard reports status for four services: Compute, Block Storage, Image Library, and Networking. Compute health, for example, is calculated based on the percentage of hypervisor hosts in a healthy state, with thresholds at 25% (warning) and 50% (unhealthy). A separate Host Health tab shows individual host status, including maintenance mode indicators.
For details on health calculation and status indicators, see the Service Health Dashboard documentation.
Exporting Metrics to an External Observability Stack
Most enterprise teams already run a centralized observability platform. PCD supports two integration methods, and the choice between them matters.
Option 1 (Recommended): Scrape from each host. Point any VictoriaMetrics/Prometheus-compatible scraper at the per-host endpoints (<host-ip>:9388/metrics for node-exporter, <host-ip>:9177/metrics for libvirt-exporter). This provides the full raw metric set, operates entirely within your network, and avoids additional load on the management plane. If you’re coming from a vROps environment where you fed host metrics into a third-party platform, this is the direct equivalent, except the endpoints are standard rather than proprietary.
PCD also supports a remote write configuration from the management plane, though this provides a reduced metric set and requires manual configuration. For details on both integration methods, see the Exporting PCD Metrics documentation.
VMware Equivalence Mapping
| VMware Concept | PCD Equivalent |
|---|---|
| vCenter performance charts | Built-in Grafana dashboards (per-region, host and VM metrics) |
| ESXTOP | Direct scrape of per-host exporter endpoints (curl <host-ip>:9388/metrics) |
| vROps / Aria Operations | Export metrics to an external observability platform for custom dashboards, alerting, and capacity planning |
| vCenter alarms | Grafana alerting, or alerts configured in an external monitoring stack |
| vSphere Health Checks | PCD Service Health Dashboard (compute, storage, image library, networking) |
Operational Considerations
No installation required. The monitoring stack is provisioned automatically per region. Metric collection starts when the hypervisor role is assigned.
15-day retention. Metrics on the management plane are retained for 15 days. Organizations requiring longer retention should export metrics to their own observability stack.
Recording rules reduce the forwarded metric set. The full metric set is available only at the per-host exporter endpoints. The management plane receives a reduced set after recording rules are applied on each host. For capacity planning or detailed troubleshooting workflows that require full fidelity, scrape the hosts directly.
Audit logging is separate. PCD also provides audit logging for API events across Identity, Compute, and Networking services. Audit logging captures who did what, when, and from where, and is covered in a separate technical bulletin.
Where Monitoring Fits in Your Evaluation
PCD’s monitoring stack covers the core visibility requirements that vSphere administrators depend on: host-level and VM-level metrics, dashboards, service health, and the ability to export everything to an external platform. The key architectural difference is that PCD exposes metrics through open, scrapeable endpoints rather than a proprietary interface, which gives your team more flexibility in how they consume and act on that data.
One thing worth considering as you plan a migration: PCD does not include a dedicated capacity planning tool. If your team currently relies on vROps for that workflow, it moves to your external observability platform. The per-host exporter endpoints provide the raw data you need. The migration task is building the dashboards and alert rules in your existing tooling, not waiting for a proprietary add-on.
For a closer look at the metrics pipeline and available metric names, start with the PCD Monitoring documentation linked in the Exporting Metrics to an External Observability Stack section. If audit logging is also on your evaluation checklist, the audit logging technical bulletin referenced above covers that topic in the same format.
