Blog / MES
MES · Guide

MES vs Historian: Signals, Events and Context

OL By OEE Lab Editorial |Updated 2 November 2026 |6 min read
Short answer

A process historian stores time-stamped tag values, often many per second, for years: speeds, temperatures, pressures, counters. An MES stores production transactions: orders, stops with causes, counts, lots and results. The historian tells you what the signal did, the MES tells you what it meant, and larger plants use them together.

Open both systems on the same seven minutes of the same line and the difference is obvious. One shows a curve that drops to zero, the other shows a line of text with a cause and an order number.

The same stop, in both systems

Example: a filling line runs order 4471. At 08:14 the speed tag falls to zero, and at 08:21 it recovers.

Historian: speed tag08:0008:1408:2108:30MES: order 4471Stop 7 min, cause: infeed jam
Same seven minutes. The curve shows the signal, and the band adds the order and the cause.

The historian can tell you how fast the speed fell and what the pressure did just before. It holds no order number unless someone sent it one.

The MES can tell you the order, the shift, the cause an operator gave and the effect on OEE. It usually does not keep the curve.

QuestionHistorianMES
When did speed reach zero?08:14:07, to the second or better08:14, as a stop start
What did the pressure do first?Yes, if the tag is collectedUsually not stored
Which order was running?Only if it is stored as a tagYes, order 4471
Why did it stop?Only a fault code, if collectedInfeed jam, from the operator
What did it do to OEE?Needs extra logicAvailability loss on that shift
One event, two records. Neither one replaces the other.

Why they are built differently

A historian is built for one shape of data: tag, time, value, quality flag. It ingests quickly, compresses hard and reads long time ranges fast.

Collection normally runs through OPC UA, older OPC or vendor interfaces, with local buffering so a network outage loses little. Compression methods such as deadband and swinging door drop points that add nothing to the curve.

An MES is transactional. It usually keeps related records such as orders, lots, stops and users in a relational database, and it enforces rules on them.

Historian: 500 tags x 86,400 s = 43,200,000 samples a day
MES: about 200 events a day (stops, orders, checks)
43,200,000 / 200 = 216,000 to 1
Example scale for one area. The two workloads differ by a factor of over 200,000, which is why they use different storage.

How data flows between them

Historian to MES: process values for a batch record, energy per order, or a counter and state signal that the MES turns into events.

MES to historian: context, such as order number, product and batch ID, written as tags or event records so engineers can filter curves by what was running.

The classic query joins both: "show the oven temperature for every batch of this product where scrap ran above 3%." The scrap and the batch windows come from the MES, and the temperatures come from the historian.

Some MES products read machines directly over PLC or OPC UA links and skip the historian. See MES PLC integration and MES OPC UA integration.

Modern products blur the edge

Many historians have event frames or batch views, and some MES products keep their own time-series store.

That overlap is useful, but check three things before you lean on it.

  • Can it store a cause? A historian can log a fault code. A cause an operator chooses, with an audit trail, is normally MES work.
  • Can it hold the order context? Tags with an order number help, but a change of order mid-shift needs clean logic.
  • Does it enforce rules? An MES can block a wrong lot or a skipped step. A historian only records.

Can you calculate OEE from a historian?

You can, with work. Formulas are the easy part, and the plant logic is the hard part.

  • Planned time. The tags do not know the shift pattern or a "no orders" period.
  • Ideal rate. It changes with each product, so you need the product signal and a table of rates. See OEE by product.
  • Stop rules. You must decide when slow running counts as a stop, as covered in stop thresholds for OEE.
  • Counter resets. Counters roll over or reset at power-up, and a query must handle both.
  • Reasons. A curve cannot say "waiting for material". An operator or a mapped fault code must, as in operator downtime logging.

Watch the sampling too. A one-minute average will hide a 20 second stop, and compression can smooth a short state change away, so collect run state on change rather than on a slow scan.

These short stops matter for OEE, as what are micro-stops explains. Check the formula itself with the OEE calculator.

Who is asking the question?

A process engineer asking why quality moved: start with the historian. Continuous processes depend on temperatures, pressures and flows over time, and a plant already on a DCS may well have one.

A supervisor asking where the shift went: start with the MES. On discrete or packaging lines, run state and counts are enough, with no need for 500 tags.

An auditor or customer asking about one batch: you need both. Regulated batch production wants process curves filtered by order and lot.

A small plant with few signals and no process analysis may need neither. A signal from a machine monitoring device and a stop log can be plenty.

Four rules before you connect them

  • Collect with a purpose. Pick the tags someone will use, then add more when a question needs them.
  • Sync the clocks. If the MES and the historian differ by a minute, joins will show the wrong curve.
  • Name tags by a standard. With 10,000 tags and no naming rule, nobody finds anything.
  • Keep time-series data out of the MES database. Writing every second of data into a transactional database will slow the system that runs your orders.
See it live in Fabrico

If you have signals but no answer to why the line stopped, the missing part is the event and its cause. Fabrico is a light MES for OEE: it reads stops, speed and counts live from PLCs or IoT sensors, and operators tag the cause at the line. See how Fabrico connects to machines.

What people ask next

Is a historian the same as SCADA?

No. SCADA monitors and supervises the process, and it shows the current state with short-term trends. A historian stores data for the long term, and SCADA is often one of its sources. Our ISA-95 levels guide draws SCADA at Level 2, while ISA's own summary lists it under Level 3, so check how your plant draws it.

Is a time-series database a historian?

A general time-series database can store the same shape of data. A process historian adds industrial collectors, buffering, compression tuned for sensors and plant-oriented tools. Which fits depends on who will maintain it.

Does an MES need a historian?

No. Many MES products collect what they need directly from machines through PLC signals or sensors. A historian earns its place when you also need detailed process curves.

How long is historian data usually kept?

It depends on the plant's needs and any regulatory rules, and many keep raw data for years because compression makes that affordable. Decide the retention per tag group, not one rule for all.

Can I send MES events to the historian?

Yes, and it is a common pattern. Writing order, product and batch IDs as tags or event records lets engineers filter trends by what was running. See MES system architecture for where this link sits.

See OEE per line and shift on your own machines

Fabrico is a light MES for OEE. It takes stops, speed and counts from PLCs or IoT sensors, with add-on AI cameras counting output where a machine has no usable signal, and operators tag the cause at the line.

Book a Fabrico demo

← More OEE Lab guides · PLC signals for OEE →

OEE essentials: OEE calculator · What is OEE · Six big losses · Availability · TEEP · Cost of downtime · OEE benchmark · Free OEE tracker · Free OEE software · How to track downtime · How to implement OEE · Light MES vs full MES