13.1 Prism Analysis: Metric Charts, Entity Charts, and Custom Reports
Key Takeaways
- Objective 4.1 asks explicitly for the difference between metric and entity charts, which is a named knowledge statement rather than incidental detail.
- A metric chart starts with one metric and plots it across several entities, which is how you compare objects against each other.
- An entity chart starts with one entity and plots several metrics for it, which is how you correlate symptoms on a single object.
- Charts are built on the Analysis dashboard in Prism Element and can be arranged into custom report configurations in Prism Central.
- Choosing the wrong chart type is the most common reason a performance investigation stalls: comparison needs a metric chart, correlation needs an entity chart.
13.1 Prism Analysis: Metric Charts, Entity Charts, and Custom Reports
Objective 4.1 includes a knowledge statement worded unusually precisely: "demonstrate understanding of the difference in Metric and Entity Charts." When a blueprint singles out a distinction like that, it is telling you the distinction will be tested.
The Analysis Dashboard
Prism Element provides an Analysis dashboard where you build charts over a chosen time range and overlay alerts and events on the same timeline. Charts are added to the dashboard and persist, so a set of charts you use regularly becomes a working view rather than something rebuilt each time.
There are exactly two kinds of chart, and they are constructed from opposite starting points.
Metric Chart: One Metric, Many Entities
A metric chart starts with the metric. You choose what you want to measure — for example controller latency, CPU usage, or IOPS — and then choose which entities to plot on it.
The result is one axis of measurement with several lines on it, one per entity. That makes a metric chart the tool for comparison:
- Which of these twelve VMs has the highest storage latency?
- Is CPU usage even across all six hosts, or is one carrying more?
- Which container is absorbing the most write IOPS?
When you build one, you are answering: "how does this one measurement differ across these objects?"
Entity Chart: One Entity, Many Metrics
An entity chart starts with the entity. You select a specific object — one VM, one host, one container — and then plot several different metrics for it on the same timeline.
That makes an entity chart the tool for correlation:
- On this one VM, do the latency spikes line up with the IOPS spikes, or do they occur when IOPS is low?
- Does this VM's CPU ready time rise at the same moment its response time degrades?
- Does memory pressure on this host coincide with the swap activity reported inside its guests?
When you build one, you are answering: "on this one object, what happens at the same time as what?"
[!IMPORTANT] The one-sentence version: a metric chart fixes the metric and varies the entity; an entity chart fixes the entity and varies the metric. Comparison → metric chart. Correlation → entity chart.
Choosing the Right One
| Investigative question | Chart type |
|---|---|
| "Which VM is the noisiest?" | Metric — plot latency across candidate VMs |
| "Why is this VM slow?" | Entity — plot its latency, IOPS, CPU, and CPU ready together |
| "Is the cluster balanced?" | Metric — plot CPU usage across all hosts |
| "What happened on host 3 at 02:00?" | Entity — plot several metrics for host 3 across that window |
A realistic investigation uses both in sequence. Start with a metric chart to find which object is anomalous, then switch to an entity chart on that object to work out why.
[!TIP] The blueprint pairs this knowledge statement with "Adding a Chart" and "Editing a Chart" in its reference list, and Objective 4.4 depends on it: recognizing CPU contention, CPU ready, memory constraints, and I/O latency from Prism analytics means building the right chart first. Section 13.3 covers what those metrics mean.
Time Range and Event Overlay
Two features apply to both chart types and are worth knowing:
- Selectable time range. Widen the window to see whether behaviour is a spike or a trend; narrow it to isolate an incident.
- Alerts and events overlaid on the timeline. This is where the analysis dashboard earns its place: seeing that a latency increase begins exactly when a node entered maintenance mode, or when a rebuild started, converts an unexplained graph into an explained one.
Custom Report Configurations
Charts answer questions interactively. Custom report configurations in Prism Central turn them into something you can hand to somebody else on a schedule.
A report configuration lets you:
- Select the entities to include, commonly by category (section 3.2) so the report follows the group rather than a fixed list.
- Choose the metrics and widgets to show.
- Set a time range.
- Schedule generation and email the output through the SMTP relay configured in section 11.1.
Typical uses: a monthly capacity report for management, a per-application-owner utilization report, or a chargeback view by business unit. Because categories drive entity selection, a report built for "all VMs in the Finance department" stays correct as VMs are added and removed — the category membership changes, the report does not have to.
Where This Sits Relative to NCM
The charting and reporting described here is native Prism functionality. NCM Intelligent Operations (section 1.3) extends it with machine-learning capacity forecasting, behavioural anomaly detection, and efficiency recommendations. For the NCA, know that the basic charts and reports come with the platform and that the predictive layer is an NCM capability.
An administrator wants to determine which of fifteen virtual machines on a cluster is experiencing the highest storage latency. Which chart type should be used?
Having identified one slow virtual machine, the administrator now wants to see whether its latency spikes coincide with its CPU ready time and its IOPS. What should be built?
What advantage comes from selecting entities by category when building a custom report configuration in Prism Central?