The full power of TDengine, now free forever for up to 5,000 tags.

Explore More

PLC Data Collection Without Overloading Controllers

Juno Qiu

November 4, 2026 /

A new historian or analytics application may need only a few PLC tags, but it joins every other client already asking the controller for data. The important question is not just whether a connection succeeds. It is whether the complete communication workload remains acceptable during normal production, maintenance and abnormal conditions.

The FactoryTalk Historian ME setup guidance explicitly says the effect of collection depends on the ControlLogix configuration and that an additional data server’s load must be considered. Treat that as a vendor-specific example of a general commissioning question, not a universal limit.

Inventory the existing requests

With the control-system owner, list HMIs, engineering workstations, historians, gateways and other clients, including requested tags, update rates, connection counts and known busy periods. Check whether an approved existing server already exposes the required information. Reuse can reduce duplicated requests in some designs, but only after its owner confirms capacity, data semantics and availability.

Separate the use cases. An hourly energy report, a maintenance trend and a fast sequence-of-events investigation need different information. A shorter collection interval does not necessarily capture a short event: the controller, source interface and collector all have their own update behavior. Agree on the signal and fidelity required before increasing traffic.

Establish a baseline before connecting

Record controller and communication diagnostics available for the specific model: task timing, CPU and communications utilization, connection counts, response times and errors. Observe representative production conditions and known peaks. Use the vendor’s diagnostic guidance; monitoring itself can add workload.

Set acceptance and rollback criteria with the control owner. Include limits for control-task timing and operational symptoms, not just a single CPU percentage. Do not copy a threshold from another model or application. Rockwell’s PlantPAx guidance discusses controller utilization and communication as part of a particular system design; its numerical recommendations are not universal.

Add a small workload and measure it

  1. Approve a read-only set of tags and a starting collection rate with the site owner. Verify that access controls actually prevent unauthorized writes where supported.
  2. Start with a bounded group during an agreed window. Compare diagnostics and application responsiveness with the baseline.
  3. Increase the tag set or rate in small steps, changing one meaningful variable at a time. Record configuration and observed impact at each step.
  4. Exercise the expected peak and an approved client disconnect or restart. Check connection recovery and whether a reconnect creates a temporary burst.
  5. Stop or roll back if agreed limits are exceeded. Investigate before continuing; a green connection icon is not acceptance.

Grouping requests or using an existing aggregation service may help in a particular implementation, but neither guarantees lower load. Request sizes, controller behavior, network paths and simultaneous clients all matter. Validate any tuning against the actual device and vendor guidance.

Example: record a controlled rollout

An engineer and the control owner agree to trial 20 read-only tags at a five-second interval during an approved operating window. These numbers are illustrative, not recommended settings. Before enabling the client, they record task execution times and missed deadlines, controller and communication utilization, connection counts, request errors and operator-interface response during a comparable production period.

They then enable the 20 tags and record the same measurements, actual request rate and any operator complaints. If the first step meets the site’s pre-agreed criteria, they add another 20 tags at the same interval and repeat the observation. For each step, the record identifies the configuration, operating conditions, observation window and measured change from baseline. Acceptance requires task timing and communication behavior to remain within the control owner’s documented limits, with no new unacceptable errors or loss of operator responsiveness.

If a task deadline is missed, an agreed limit is crossed or control or operator responsiveness deteriorates beyond the agreed threshold, the team stops the test and removes the added workload using the approved rollback procedure. They record the trigger and post-rollback measurements, then investigate with the control owner before trying another configuration. The site chooses the actual thresholds and test conditions for its controller and process.

Make the workload supportable

Document who can add tags, how new clients are reviewed, what diagnostics are monitored and who responds to sustained errors. Repeat the assessment after a meaningful configuration change. The target is an approved data service that remains compatible with the controller’s primary job: operating the process reliably.

Related Articles