PI System Monitoring Templates (PSM): How to Monitor the Health of Your PI Infrastructure

Process monitoring is not enough. PI administrators also need to monitor the systems that collect, store, calculate, and serve the process data.

The PI System Status Monitoring tools and PI System Monitoring templates can help teams represent PI infrastructure as assets and monitor key health signals.

What PSM is for

A PI environment can include:

• PI Data Archive servers

• PI Asset Framework servers

• PI Analysis services

• Interfaces and connectors

• Supporting Windows services and network paths

A failure in one component can first appear as a missing value, a slow analysis, or a bad dashboard. Infrastructure monitoring helps the administrator detect the cause earlier.

Start with the critical data path

Do not monitor every available metric on day one. Start with the components that can interrupt data collection or data delivery.

For each critical component, define:

• What must be running

• What normal behavior looks like

• What threshold requires attention

• Who owns the response

This keeps monitoring useful and limits alert noise.

Monitor PI Data Archive health

Useful checks can include:

• Service availability

• Archive and disk capacity

• Event queue or buffering conditions

• Snapshot and archive update behavior

• Client and interface connectivity

• Resource use that remains outside the normal baseline

A single high CPU value does not always indicate a PI problem. Use trends and known baselines before you create an alert.

Monitor interfaces and connectors

An interface can remain online while individual data streams stop updating. For this reason, service status alone is not sufficient.

Monitor both the interface and the data it sends. Useful checks include:

• Interface or connector availability

• Last successful data update

• Data gaps

• Communication errors

• Buffering conditions

• A sudden change in the number of active points

The exact checks depend on the interface type and architecture.

Monitor AF and analysis services

For PI Asset Framework and PI Analysis Service, monitor conditions that affect calculation and model reliability.

Review:

• Service availability

• Analysis error and warning states

• Calculation lag

• Skipped calculations

• Queue or workload indicators

• Repeated failures by template or asset

A service-level metric tells you that a problem exists. Asset and analysis context helps you find the calculations that need attention.

Use templates to standardize monitoring

Templates make monitoring easier to repeat across servers, sites, and environments.

A good template should use consistent attribute names, units, limits, and ownership. Do not modify standard templates without documenting the change. If you extend a template, make the extension clear so another administrator can understand it later.

Set thresholds from normal behavior

Static thresholds are useful for clear limits such as free disk space. They are less useful for metrics that change with workload.

For workload metrics, first establish a normal baseline. Then alert on conditions that are sustained, abnormal, and operationally relevant.

This approach reduces nuisance alerts.

Connect infrastructure health to data health

Infrastructure monitoring and data-quality monitoring solve different problems.

Infrastructure monitoring tells you whether services, interfaces, and servers are healthy. Data-quality monitoring tells you whether expected signals are current, valid, and useful.

Use both. A service can be online while a critical tag is stale. A tag can also be healthy while an analysis or display uses the wrong source.

Recommended operating practice

Review PI infrastructure health continuously, not only during incidents. Keep the monitoring model current when servers, interfaces, and services change.

A practical monitoring program should help the PI team answer three questions quickly:

• What failed?

• Which data or users are affected?

• What changed before the failure?

When the monitoring system can support these questions, it becomes a useful operating tool instead of another dashboard.