What is the practical scale-ability for adding devices on a single license?

Hey everyone!
Any advice on this would be appreciated.

I understand that the unlimited license allows for an unlimited amount of devices to be added, but how practical would it be to add, say for example, 10,000 devices (with approximately 100 tags each)? How do we determine the physical hardware requirements to scale an application like this up? Or would it not even be feasible and require multiple gateways?

Sorry, we don’t have any published guidelines or even any internal benchmarks for this kind of thing. It’s also going to vary depending on which driver you are using.

10,000 devices is an order of magnitude more than I would recommend. We’ve seen a couple hundred up to maybe a thousand, but the current architecture of the drivers doesn’t lend itself to many more.

There’s also an issue of how to configure the devices in bulk, both initially, and to when it comes to maintenance/modifications.

In the future I this will change, starting with the Modbus Driver, but the timeframe for these changes to start taking place is something like 8.1.


I am doing a university project where I need to setup multiple HMI/SCADA systems with a large tag base. 10000 devices with 100 tags each is around the size ev of the system I need to setup. Each tag will be read via OPC DA ,UA or Modbus and will update very 1 minute on average. I also need to log all this data to a historian. My server specs below,

  • 16 gb ram
  • 1 TB SSD
  • Windows 10 OS
  • i7 -7700 HQ processor 2.8 Ghz

Do you have any guidelines about how many ignition servers I need to setup this scenario? There are some very clear bench marking with some of the other products I am testing with.


You might be able to get away with 4 servers… not sure. That would still be 2500 devices / 250,000 tags per server.

There are people out there with more tags on one server than that, but the high device count is likely going to be an issue. Ignition is not well suited for having thousands of devices per gateway right now.

I can scale down on the number of devices for my test at least, and it will be simulated. However, the system needs to have a million tags. Also ,there will be 200 web clients. If there are four servers, how can I make sure that the users can access the graphics with the same IP address? Is there a way to setup a front-end visualization server that that communicates to the four load balancing servers?

Any architectures/documentation you can point me to will be much appreciated.


I can’t link directly to it, but look at the “Scale-Out” diagram on this page: https://inductiveautomation.com/ignition/architectures

I would suggest working with an intermediate OPC server (like kepware), but apparenlty kepware limits itself to only 256 devices (https://www.kepware.com/en-us/support/knowledge-base/2010/does-kepserverex-limit-channels-devices-or-tags/ )

At least in my case, I can set it up with a single device, but not 100% sure how to architect a million tag system with Ignition. Looked at the scale-out architecture, but still not clear how many tags I can put through a single I/O gateway.

There’s no limits in the licensing, it’s going to come down to the kinds of devices you have, how frequently tags are changing, historizing, client count, a million other things, and probably most importantly the hardware you have available to run each IO gateway on.

You’ll just have to try it and see. There isn’t a calculator you can plug numbers into and get an answer.

Thanks Kevin, I appreciate your input.

This is for a class project, so, I will have a simulation OPC UA server generating 1 million tags, updating every minute on average. The entire million tags will be historized and 400k tags will be alarmed on. There will also be 200 concurrent web clients. The hardware specs on each I/O gateway is below,

  • 16 gb ram
  • 1 TB SSD
  • Windows 10 OS
  • i7 -7700 HQ processor 2.8 Ghz

As I said, I appreciate your input.