ometimes I know I'm just asking for it. I know there are subjects for this column that, if I write about them, could generate a level of emotional response that dwarfs any more-objective reactions that I'll get. My lineage will be questioned, inbreeding might be implied, and, in general, my very fitness to hold this job will be called into question.
I know all this. I can't help it, though.
It's about Linux, you see.
Our editors periodically ask around about the state of Linux affairs in machine control or in general industrial automation. This month we're publishing responses to a reader question posed in March asking if Linux is finally ready for machine control and if so, where's the evidence of that.
I'd hoped we'd get some responses that began to settle issues about general acceptance, ease of use, even whether or not it's cost-effective.
We've printed a representative sampling on page 61 of what we heard back from users and suppliers.
The resounding conclusion about Linux seems to be: still not ready to be trusted and really tried, or admitted to. Or, to head off you conspiracy theorists at the pass, the major automation suppliers still want to keep it under their (red?) hat, so to speak.
It doesn't much matter why, because,to answer the reader's question,right now the evidence of genuine industrial machine control deployment just isn't very compelling.
It's not to say that Linux isn't doing well overall,it is. Novell and Red Hat recently reported quite good financial numbers, and companies such as PeopleSoft, IBM and others are pushing a very optimistic enterprise-level message for Linux. Linux is found in many, many web servers, hand-held devices, and even in the Mars Rover.
So, at the machine level,even where you might not need real-time functionality,is the lack of interest because many of you only recently have migrated away from proprietary operating systems and communications protocols to open-standards-based platforms? That was a mixed bag of opportunities and problems. You don't see the levels of support for Linux out there to help you,not just with the machine, but with ever-increasing levels of data movement requirements, as well?
I think I understand that. The OS revolutionaries point to Ethernet as a force that stormed out of the front office into the factory and took business from proprietary fieldbus supplied by some well-established automation heavyweights. They say Linux will win in a similar fashion. They said the big ol' automation suppliers had to come around to an Ethernet flavor of their own to stay in the game.
Well, not exactly. The difference is that Ethernet was everywhere in the front office. It came with an army of IT folks who may not have known "factory," but they knew the technology. There just aren't that many of your customers who have a Linux propeller head named Neo running the enterprise IT shop. So, who leads the Linux charge?
Seems like most of you believe the best thing to do is to stay abreast of your customers' open-architecture demands and see where they lead you. There are manufacturing companies out there playing with Linux in many applications. DaimlerChrysler is using Linux as the OS for onboard GPS and web-browsing in some Mercedes. That's not a welding robot train, but success in one area will mean curiosity about others.
I'd really like to see it gain factory-floor inertia as a result of one or two compelling first-hand success stories. The competition would be good. We would be delighted to report them.