Tag Name based communication

J

James Ingraham

@Curt: Also witness the tremendous fortunes made off of proprietary technology.

Here's a partial list of Fortune 100 tech companies:

7 AT&T
10 Hewlett-Packard
20 IBM
36 Microsoft
38 Dell
56 Apple
58 Cisco Systems
62 Intel
63 Oracle (post Sun, approx)
100 Amazon.com

See anybody who made a fortune on non-proprietary products? Now granted, some of these support open standards, and Dell and HP certainly have to use standards like PCI / PCI Express, ATX, USB, etc. But the big money is definitely in proprietary systems. Even Google, at 102, makes its big bucks from its highly secret search technology. Red Hat has learned how to monetize open systems, but they would need to grow by at least a factor of 5 to reach the Fortune 500, much less compete with Microsoft.

A special note about Xerox at 152. They invented Ethernet, the GUI, object-oriented programming, the mouse, desktop publishing, bitmap graphics, the integrated development environment, and the WYSIWYG concept. All of these are now "standards." They made money off of NONE of these things. (They also invented the laser printer, and that has been a source of income, but they have never done particularly well in the basic desktop printer space.)

My company used to program our machines in C, an open standard known by millions of developers. Our customers hated it, and we switched to Allen-Bradley PLCs to make them happy.

Profibus and DeviceNet may slowly be going away, but they have to count as commercial successes. EtherNet/IP and Profinet dominate, despite the ease and openness of Modbus/TCP. Profinet and OPC (both of which again must be considered sucesses) are both based on DCOM, about as closed and proprietary as a protocol can get. CIFS has done just fine. Even though IPX/SPX is dead now, killed by an open standard, it was a success for many years.

And don't forget the dead open systems, either. For every TCP/IP or Linux, there are hundreds of systems that DIDN'T make it.

We're WAY off topic at this point, but I think it has been an interesting tangent. Personally, I really prefer open standards. At a business level, I am much more practical.

-James Ingraham
Sage Automation, Inc.
 
C

curt wuollet

Hi James

Many of those have had their fingers slapped for their zeal. It is an interesting tangent. The underlying issue is bundling. It seems one product can still offer so much perceived value that it can carry sub optimal or even unabashedly exploitative cargo like the old Trojan Horse. Like the I-phone that is so cool, people don't mind being ripped off by ATT. Like Windows that is so familiar and easy that it simply doesn't matter that it's a security nightmare. I think that is what fascinates me. That may be a bit extreme as many of the things we talk about aren't horribly detrimental and often do provide functionality, but they tend to be very expensive and often quite aggravating, long term. This trend only goes so far as in a mature market, consumers begin to want the good with the good rather than accepting totally one sided licenses and marriage as a condition of meeting. Why hasn't this come about in the fairly technical field of automation? People will argue about the most petty pros and cons of one horrifically limiting package over another while completely ignoring that both seek to enslave you and put you over the barrel. It's sort of like rebranding Heroin as a new wonder pain reliever, with fewer side effects than the current crop and arguing the effectiveness pro and con, while neglecting that common problem that they are all addictive as Hell. Fascinating.

Regards
cww
 
In reply to James Ingraham: You said "See anybody who made a fortune on non-proprietary products?". Well, yes, I see quite a few. HP, IBM, Dell, Oracle, and Amazon for starters. Even quite a bit of Apple's OS is open source software that they've re-branded. I'm not aware of any "Cisco network protocol"; they simply produce network hardware that supports open standards (open as in you don't need anyone's permission to use it). Intel's attempt at establishing a new line of CPUs (commonly known as the "Itanic") that wasn't compatible with anyone else's was a complete flop and it's only a matter of time before they pull it off life support. As for CIFS, yes it has been quite successful in it's niche - after they moved it off their own underlying protocol and switched to TCP/IP like everyone else.

So where's Burroughs, ICL, DEC, NCR, etc. on your list? The companies that, along with IBM, used to dominate the business, designing their own hardware, running their own OS, and programmed with their own languages? With the exception of IBM, they're gone, swallowed up by someone else, or living a marginal existence on a handful of legacy customers. What happened is the foundations of their businesses became commodities, and unlike IBM they didn't adapt and move on.

What you're missing is that the business isn't about selling the same old thing but giving it away for free and somehow making money off that ("we'll make it up on volume!"). What happened is that while at one time being able to put chips together, or writing an OS (or a database, or a compiler) were once cutting edge technologies. Well, they're not cutting edge anymore, lots of people can do it. They're commodities, and the thing about the commodity businesses is that prices get squeezed down to the minimum. For software, the minimum tends to be zero. If you want to make money in that type of market you've got to move on to products or services that aren't commodities rather than standing there twirling your moustache and chortling over all the customers who are unable to escape the clutches of your legacy products. Finding a successful business strategy in that sort of environment isn't easy, but then making money in any business seldom is.

I look at the major PLC vendors today, and I see a lot of little NCRs and ICLs. Companies selling expensive vertical solutions that don't work with anything from anyone else unless you buy a special gateway. Apple can get away with selling overpriced fashion accessories, but I can't see that strategy working in the industrial field.

I think that companies who make networked I/O, servo drives, and systems like that will be fine. Companies selling high end DCS and SCADA systems will probably be fine, provided they follow a "Red Hat" strategy and base their business on providing and supporting a complete platform. Companies who rely on having an existing base of customers who can't easily switch are going to find their customer bases fading away and new customers hard to get.
 
J

James Ingraham

HP didn't make their fortune with truly open systems. Yes, HP (and Compaq before them) sell "standard" PCs. But that's not the big bucks. Medical equipment, test & measurement, calculators, etc. helped put HP where they are. And don't forget HP-UX, PA-RISC, and other proprietary products that helped build the company. Oh, and they were partners with Intel on the Itanic debacle.

IBM is even worse. AIX is still kicking. Most of their stuff is proprietary. They're a big backer of open source, but they're in the Fortune 500 for gigantic systems glued together with their in-house expertise. Sure, sure, they'll run a 1000 Linux virtual machines on their big iron, which helps them sell the big iron. But the big iron is THEIRS.

And Oracle? Oracle?!?! They're the WORST. Yes, they provide some open source software. It's not even CLOSE to a money maker for them. Their money comes from big systems running big proprietary Oracle software systems. So proprietary that Larry Ellison once said, "If your business model doesn't fit our software, change your business model."

Amazon is no better. Look at the Kindle. Proprietary format. BARELY supports PDF. Doesn't support EPUB. All of their really cool technology is locked up behind closed doors. Yes, they use some open technologies that help them run their business. But they never made a dime off of open tech; they sell STUFF. And if you ask them to let you see their internal software for search, logistics, warehouse management, order handling, etc., they will tell you an emphatic NO.

As for Dell, I grant that they do mostly "standard" stuff, even though Windows and x86 aren't "standards" so much as "common." Still, they often sell non-standard form factors and screwy power supplies that work only with their motherboards. And while they've benefited from things like the ATX standard, they haven't really been a contributor to open technologies the way IBM or Google has. Still, if I have to concede one of these it would be this one.

"Even quite a bit of Apple's OS is open source software that they've re-branded."

Sort of irrelevant. They may have gotten a leg up by using the MACH kernel, but their model is to close everything down. Sure, Mac OS will run on standard x86 hardware... but it's locked down so it's difficult, with a nice little legal club over your head if you try. Their fortune has come from their proprietary features, particularly with the iPod / iPhone.

"I'm not aware of any "Cisco network protocol"; they simply produce network hardware that supports open standards."

Ha! Cisco has all kinds of proprietary extensions. They have their own OS (actually several, one of which happened to be called IOS, which Apple then had to license.) Sure, their equipment handles open networks, but they lock down everything they possibly can. Cisco's big money comes from BIG systems, and trust me, you can't just pull out a AS9000 router and drop in a Juniper T series router. Hell, even the 48-port managed switches here at my company (which are Dell branded Cisco switches) have proprietary features. Don't get me started on their VoIP phones, either.

The Itanic may have flopped, but remember that the x86 architecture is as non-open as it gets. Invented by one company, held on to by them for as long as possible, with lawsuits galore whenever they can get away with it. They forced nVidia out of the chipset business, for goodness sake. When AMD came out with a very nice set of 64-bit extensions that everyone praised, Intel waited as long as possible and then grudgingly added them... with just enough difference to annoy every compiler writer on the planet.

"As for CIFS, yes it has been quite successful in it's niche - after they moved it off their own underlying protocol and switched to TCP/IP like everyone else."

This an example of the worst of both worlds. Microsoft can say "See! We support open standards like TCP/IP!" which is technically correct but useless in practice, since you're stuck using proprietary stuff like CIFS and Active Directory and Exchange anyway.

"Burroughs, ICL, DEC, NCR, etc..."

Good point. You can add a lot of others, too. (Although, don't forget DEC was bought by Compaq which is now part of HP, again reinforcing the idea that HP made its fortune of proprietary products.) But where is the Microsoft of the open world? Of all the really important open technologies we use, did any of their inventors go from a garage to billionaire? I don't see Bjarne Stroustrup or Linus Torvalds selling their 450ft yachts because they want to get a new one bigger than Paul Allen's or Larry Ellison's.

"Companies who rely on having an existing base of customers who can't easily switch are going to find their customer bases fading away and new customers hard to get."

While I hope this is true, I don't yet see any evidence of it. I have yet to have a customer ask me for a PLC other than A-B or Siemens (and many years ago Modicon, but nobody's asked us for them in nearly a decade.) I have yet to have a customer ask me for a drawing in a format other than DWG. I have yet to have a customer or vendor send me a document in anything other than an Office format. (Okay, okay, I have seen lots of PDF, but PDF got big when they were proprietary.) I once did a Profibus network layout in OpenOffice.org Draw, purely because our electrical design team was bogged down and I suck at AutoCAD, plus it LOOKED better in Draw. I had to give it to the customer in PDF format, of course.

Would I prefer to see open standards win? Yes. Do I think they always do? No. Do I think it would be a good business decision to ban non-proprietary products from our company? No. We wouldn't last a day. And what would you do for computing, btw? x86 is closed. SPARC is open, for all that that's worth. You'd have a hell of a time outfitting your company wit SPARC based desktops and laptops. I can't think of any truly open hardware systems at the level of even most smartphones.

At home, I use a lot of open source software, and I picked Android over iPhone. But I still run Windows, because I play games, and that's really the only choice. Oh, and my wife refuses to use anything else. Would I trade my marriage for running Linux? No. And I won't trade my business, either.

-James Ingraham
Sage Automation, Inc.
 
Curt,

You mentioned patents...

I'm not aware of any patents that pertain to EtherNet/IP.

Do you know specifically of any patents held by ODVA or Rockwell that cover EtherNet/IP?
 
L

Lynn August Linse

>In reply to Lynn August Linse: Are you
>saying that ODVA does not require
>conformance testing, or are you saying
>that they are simply not currently
>pursuing anyone who distributes hardware
>or software without it?

I don't particularly worry about such things - have any of you read your cellular "Terms & Service Agreement?" How about your Internet access? Most of what you do daily is against the rules, because it gives the provider the right to close your account if you cause problems.

In the town I live in, I don't have the right to park my car in the same place for more than 72 hours - even on my own property.

Until the early 2000's, one could not use MODBUS without submitting all documents containing the word MODBUS for approval by Schneider Electric, plus you had to submit copies of all sales records of MODBUS products to Schneider Electric annually.

In 1998, how many of you submitted your sales records of all MODBUS products as required?

Why is Modbus so popular given you were required to disclose all of your sales data to the owners of the MODBUS specification?

It is an interesting tidbit of history. Had everyone worried so much about the rules, no one but Schneider Electric would be using MODBUS today :)
 
In reply to Lynn August Linse: When it comes to industrial communications, you undoubtedly know more than anyone else here, and certainly far more than I do. I think though I realize that the question I asked though was probably not one that you could reasonably give a definite answer to.

There are certain things however that companies can and cannot do with respect to trademarks. Despite Schneider Electric having once *asked* people to give them copies of documents which contain the word "Modbus", unless you were using their trademark in an infringing manner you could simply have told them to take a flying leap. Simply using a trademarked word doesn't infringe upon it. If it did, Control.com couldn't exist as we mention trademarks here all the time. You have to use the trademark in an infringing manner to have a problem, and simply using a word or phrase doesn't necessarily do so.

With Ethernet/IP, you can either just be very careful as to how you use the phrase, or as you said avoid the problem by simply not using it at all and saying that it "works with AB PLC model xyz". I would be more than happy with that solution.

AB / ODVA spends a lot of effort telling people they *must* license Ethernet/IP from them, but they never really say why. They waffle about patents, but I will be honest and say that I haven't seen any direct evidence that they have any patents on anything about Ethernet/IP that would interest me (explicit messaging). In fact it is quite possible that there is nothing in there that is even patent-able. A straightforward client - server protocol over TCP/IP wouldn't be. They may have something with regards to their safety or drives versions of their protocol, but those don't really interest me (or for that matter many other people).

They do make you sign a contract to see their specs, but that puts you under contract law. If you avoid their specs, then you avoid the contract trap. That's a very strong reason for me to *not* sign their contract. A company like yours would be in a very different position however, as you would likely want ODVA certification anyway for marketing reasons.

I just had a look at Wireshark and rather interestingly enough, it has a decoder for Ethernet/IP as well as several other protocols (such as Ethercat, Powerlink, and Profinet). There's still the big step of going from a decoder to a driver, but at least that doesn't involve starting from scratch. This is not something that I can follow up on at this time, but you have certainly provided a lot of food for thought.

I would like to however clarify what I think is the essential difference between a protocol that is "open" and one that is "not open". "Open" means that anyone can implement it without restriction. You don't have to sign any contracts, license any patents, or undergo any certification. You just go ahead and do what you want. Furthermore, *and very importantly*, all those rights are passed on undiminished to anyone who uses it. If you can't do all of those, then it's not open. It's not a matter of price, it's a matter of the freedom to do things.

There are lots of closed protocols that have been decoded. That doesn't make them open. There are lots of closed protocols that have been licensed to third parties. That doesn't make them open either. When I say "open", I means "open like the Internet" where you do whatever you want, not "open like the cable company", where you get what the cable company will allow you have.
 
In reply to James Ingraham: In nit picking over details you seem to have overlooked the gist of what I said. Have a look at what IT customers who aren't maintaining legacy systems are buying from those companies today. They're not buying PA-RISC with HP-UX anymore. They're buying commodity servers with Red Hat or SUSE. By and large, the OS, database, or web server are commodities. If you're not happy with the price or service you are getting from Dell, you can switch your future orders to HP or IBM with much less pain than would have been the case in the past.

What happened in the IT market was these companies recognized that the lower levels of the "stack" have become, or are becoming commodities. The technologies are mature and improvements are incremental. If they wanted to prosper, they had to change strategy. HP's server business competes head to head with Dell on price, delivery, reliability, and service. They don't rely anymore on the fact that you once bought an HP server and are therefore stuck with them as a supplier forever more.

It's not that these vendors *wanted* to have a competitive market. The customers forced it on them by changing their buying habits. Fifteen years ago most people in the IT business would have laughed if you suggested this would ever happen. Ten years ago they would have been doubtful. Five years ago it became common knowledge. Today it's accepted practice. I could go over your previous response point by point, but really you're looking on the negative side in each case to try say the glass is half empty instead of saying it's half full (by the way, you do know that Amazon is also a major IT industry service supplier?). Of the companies that you mentioned, here's their ranking in terms of how much work they do to write Linux (as of last summer): Intel (#2), IBM (#4), AMD (#7), Oracle (#8), Google (#12). I can assure you they aren't spending all this money out of the goodness of their hearts.

Will this same process happen in the automation business? I don't see why not. Let's look at the following:

1) Ethernet with TCP/IP (or UDP/IP) has already displaced a lot of the proprietary networking hardware. Commoditisation has already started in that area (although many of the application protocols themselves are still a problem).

2) There's been no significant innovation in PLC languages in the past ten or more years. It's a mature market and ripe for commoditisation.

3) Distribution channels are becoming loosened as local dealers close down and the Internet becomes a more common medium for doing business.

4) The people using PLCs these days are much more computer literate, and are more attuned to the idea of control systems doing more than what typical PLCs are capable of these days.

5) Standardized embedded hardware designs are available, and they almost always use Linux, which means that application development in much easier than in older systems. That lowers the barriers to entry.

6) I/O systems are moving out of racks and onto networks, which breaks the link between PLCs CPUs and I/O.

So how would a vendor make money in this type of market? They can still sell I/O, servo drives, sensors, etc. Why would they want to do this? If you're one of the existing major vendors, you probably wouldn't want to. If your main business is I/O however, you would be quite happy to undermine the major vendors as that would open up a bigger share of the I/O market to you. If your business is stuck in a niche because the established vendors have a lock on the market, the classic business strategy is to do something that disrupts the existing market to create new opportunities to exploit.

Who else would benefit? Well, I know of several large system integrators who have their own soft logic systems. They like their own system because they can tune the features to how their equipment usually works and so are able to put systems together with much less programming. Customers tend to resist buying them however, because they're still proprietary systems. If a customer is going to buy a proprietary system, they may as well buy a third party system and at least have some vendor independence at the integrator level. A common open source system would remove that objection while still providing most of the advantages that the integrators really want.

What would I suggest that automation system designers do? Well for starters, I would suggest that they supply whatever it is that their customers want. Your customers are adults and they can make their own decisions. Your customer wants an AB PLC? Well fine, that's up to them and I am quite sure that AB makes very nice conventional PLCs.

On the other hand, people in the automation business should keep abreast of technology and educate themselves as to what can be done. Make contact with people in your area who can do custom applications, web servers, databases, etc. on open systems and exchange business cards with them. When a customer application comes up that you think could fit that sort of thing, you have contacts lined up that can work with you. Try things out yourself, so you've got your feet wet already when you want to take on something yourself.

The problem with waiting for your customers to ask for something, is that a lot of the time they aren't going to ask you for it if they don't know you can do it. They're going to go to someone else. Many times, the customer won't ask because they don't think there is anyone who *can* do it.

I said "you" above, but that's the third person plural "you" rather than the second person singular "you". It really applies to anyone in the business. Let's take your company for example though. I just had a look at your web site. It says you integrate robots with AB PLCs, mainly for material handling systems. You said you're happy with AB PLCs, and you're happy with the robots you're using, so let's forget those (including the programming software). What aren't you happy with? What would you like to be able to offer your customers, but you can't because it doesn't exist, it doesn't do what you want it to, or it's simply too expensive? That's where you (and anyone else) would start. If you say that you don't have any problems and you already live in the best of all possible worlds, well, congratulations. There's not too many people who can say that however.
 
C

curt wuollet

No, I don't. But does that mean the company that tried to patent any use of PLCs with browsers, etc etc. doesn't have some obscure patent on the technology? The latest game is to get your stuff rolled into a standard and then whip out the patent card. But, like I mentioned, the control they do retain should be sufficient to prevent widespread adoption. It makes one wonder if they have seriously considered the effect on the market that a truly Open protocol could have if adopted by the major players.

Regards
cww
 
J

James Ingraham

@M Griffin: I believe I understand your point.

Let me reiterate mine. If you want to be a gazillionaire, you have to have a secret sauce.

Yes, Red Hat makes money. Yes, some big name companies have been forced into a commodities market where before they had high margins.

If you want to hit it big, you have to have something nobody can copy. HP has much higher margins on their printer ink than on their servers. Apple; 'nuff said. Microsoft; 'nuff said.

Could the mighty automation suppliers be brought low the way some of the big tech companies were? Sure, it could happen. We can even hope for it. There's nothing flawed in your vision. But you semi-alluded to the fact that to compete in that world you'd have to have SOMETHING ELSE giving you your margins. That's the secret sauce I'm talking about.

"I would suggest that they supply whatever it is that their customers want."

I'm glad we agree on that fundamental point.

"On the other hand, people in the automation business should keep abreast of technology and educate themselves as to what can be done."

In my opinion, this is true for anyone in any field.

"The problem with waiting for your customers to ask for something, is that a lot of the time they aren't going to ask you for it if they don't know you can do it."

A classic problem. I really don't know how to handle it.

One final point. In about 1999, I predicted the PLC would be replaced by PCs within 20 years. It didn't take me twenty years to figure out I was wrong. One fundamental thing I didn't take in to account was that the PLC guys would fight back. Now we've got motion instructions in ladder logic, Ethernet connectivity, SD card slots, etc. There's nothing that prevents a PLC from having direct SQL calls in ladder, or whatever. Of course, the argument goes that these are really P *A* Cs, now, rather than PLCs. So maybe I was right after all. :)

-James Ingraham
Sage Automation, Inc
 
Top