The Internet of Things isn’t a fad. It would be hard to argue that Microsoft, Cisco, IBM, GE and Dell (which is committing ONE BILLION dollars) are all caught up on a passing fad. But in the same vein for us in industrial automation, IoT isn’t really any different than anything else we’ve been doing for the last forty years.
We’ve always been moving data to Enterprise systems. Back thirty years ago, machine operators would have daily production tally sheets to fill out. Machines had mechanical counters. The operators would record the value of those counters and turn them in at the end of the shift, reporting key values like production, scrap and quality. Those sheets would go to the “IBM room” – always called that because it always had an IBM mainframe in it – where a keypunch operator would keypunch those values onto punched cards. Sometime later that day, those cards would be loaded up and transferred to a huge magnetic tape roll. Sometime after that, the tape would be mounted again to print daily, weekly, monthly and yearly reports that someone hand delivered to various offices around the plant. Other copies would be mailed to corporate for analysis days or weeks later.
If this sounds archaic to you or something from the Slate Rock and Gravel company – where Fred Flintstone worked – I wouldn’t argue with you. But that’s exactly what moving data from the factory was like for almost fifty years.
Eventually we realized that having humans write things down and other humans keypunch it onto punched cards and other humans load those cards into a computer was really labor intensive. So we endeavored to build systems to do that. The first ones replaced the paper with floppy disks. Operators entered data into a computer and the data was stored on a floppy disk. The disk was then carried to some other computer and entered into the Enterprise system.
As time went on, we started to do direct communication between the factory floor controllers, which were now more sophisticated, and the Enterprise systems, which were also more sophisticated. Unfortunately, there were all sorts of hurdles to overcome: differences in baud rates (yes, it was serial), encoding standards (see my three-part article on ASCII encoding), data formats and much more. It was a herculean effort to link these systems and it could only be done by very experienced programmers working for a long time for companies that had the funds to make that kind of integration happen – the P&Gs and GMs of the world.
It got easier with the invention of OPC (New called OPC Classic) but not all that easy. OPC made the data collection from the machine controller easier, but there were still hurdles. You had to move the data from the factory floor computer running OPC drivers to the computers running the Enterprise. We still had the same issues as before: data encoding, data formatting, naming, the loss of meta-data along the way and more. At least now we were using Ethernet so it wasn’t so much of a hassle to physically connect things.
Now with IoT technologies it’s easier than ever before. We have a plethora of media to use like Ethernet, WiFi, cellular, Bluetooth and wireless mesh technologies. We have new protocols to move data like MQTT and AMQP. We have Cloud based databases and all sorts of tools for creating applications running on our servers or in the Cloud. We have various ways of securing data as that is now a huge concern for everyone in Industrial Automation. We have totally new architectures like OPC UA.
Even though it is much easier now to move data to the Cloud and to Enterprise systems, that doesn’t mean it’s easy to do it correctly. There are still pitfalls and traps – ways to really screw it up and that’s what I’ll discuss in the next article.