Ignition on Cellular PC Logging to Azure Database

Hello all,
I am working on trying to get data from some of our customer machines. We manufacture machines that are then taken to customer sites and we’re looking to use ignition as a part of a service package for remote monitoring.

My thought as of now is to use one cellular PC per customer site with the SQL bridge module to log to an Azure database. Just wondering if that approach seems ok and if there are any other suggestions?

We need to try and stay off the customer network so that’s why we’re going cellular.

Thanks for your help

That sounds tailor-made for Ignition Edge MQTT.

I was reading about edge MQTT and it has to use a broker. I know that azure has the IOT Hub which is a broker but I think it’s little funky. do you have any experience using edge?

Just playing in my lab. /-: (My customers don’t have lot’s of little remotes.)

But yes, Edge MQTT needs a broker, and it really ought to support the sparkplug architecture. An Ignition instance in the cloud could be your broker up to 250 remotes, but for many remotes you’d want a dedicated broker like Cirrus Link’s Chariot server.

For what you describe, MQTT sounds like the right solution. We went a more manual, but cheaper route and are using DIGI Connect Sensor cellular devices (about $400). They take 4-20mA, 0-5V or fetch data via Modbus and log data to DIGI’s cloud, but it was easy to write a script that does an httpGet to the DIGI cloud. The results are returned in JSON format to the Gateway script, the data is pulled out and tucked into each associated Ignition tag. We use an Azure DB for our Ignition historian, and Ignition itself is running on an Azure VM. The advantage is that there’s no computer (or Edge license) required on site, but they only have 4 analog inputs - so you’re REALLY limited, unless you can go Modbus.
That said, we only fetch the data once an hour for historical purposes, and even if we are using it for control, our process is so slow that it allows us to take a reading every fifteen minutes. If you need faster sampling, there is certainly a lower limit as doing the httpGet and parsing it takes some tens of seconds, plus you’ll have to pay the cell data bill!