# UNIX and VMS based systems V.S.. NT based systems.?

V

#### Victor Cedillo

" List ",
I'm interested in your thoughts regarding the direction of NT and ethernet
SCADAs versus UNIX or VMS based systems.

Problem with, future of etc...

Victor Cedillo
[email protected]

D

#### Davis Gentry

Uh oh..... and thus from one question the great war begins anew.

Davis Gentry

J

#### John O'Beck

IMO
Unix is still very viable, but losing popularity.
Like it or not, the world is going NT.

R

#### Ranjan Acharya

As a systems integrator we are seeing a definite shift towards NT versus
UNIX (all flavours) and VMS. Wonderware, Siemens PCS-7 (WinCC), A-B RSView,
CiTect and so on are all Windows NT only.

I think that the List has clobbered the Unix versus NT thing. For us,
supporting all the flavours of Unix is a real hassle, however, Unix seems
more robust. From what we have seen / read / heard, Linux is not quite
there yet compared to QNX, AIX, Solaris, HP-UX and so on.

I have not worked on VMS since the early 1990's, so I cannot comment on

As far as Ethernet goes, we see Field Buses such as ControlNet, PROFIBUS,
Foundation Fieldbus and so on still being useful (and sometimes better)
alternatives to Ethernet. Ethernet may win in the end (when it reaches
1Gbps with some sort of true determinism), but we are still seeing
tremendous support for OEM-engineered buses -- depending on whose SCADA
platform you choose.

Ranjan Acharya
Grantek Control Systems http://www.grantek.com/

M

#### Mark Hutton

Linux is actually rapidly gaining in popularity and stability. While it is
still some way from becoming a mainstream enterprise product, mainly due to
the lack of support skills, I wouldn't expect it to stay that way.

My experience of Linux is that it is both more stable and at least as easy
as Windows 98 to install (that's despite my version coming on six CD's).
Suprisingly, I found Linux easier with hardware too; it took me four time as
long to install my ATI All-in-Wonder 128 in Win98 than it did in Linux
(which according to the box doesn't support Rage 128). That's despite a trip
to the Suse web site to download a driver. I also have an unbranded network
adapter card which despite coming with a driver disk took me about ten
attempts to install and half a dozen reboots in windows. In Linux the
installation of the network adapter was about two minutes, PROBE picked out
the cards chipset, and YAST insatlled it, no problems.

Of course Windows 98 is not Windows NT, and as I have limited experience of
NT I can make no comment.

There are already some good SCADA packages that run on Linux, notably PCCs
Aprol and M&Rs AutomationX. I also think that the change to Windows 2000 is
going to cause Windows (and us) more than a few problems.

Regards
Mark

P

#### Peter Clout

I have to disagree with the forcefulness of John O'Beck's statements in
reply to the above. In my view:

VMS
This is a great mission-critical operating system with a solid technical
foundation and proven in countless mission critical systems. The momentum
will keep it running. Compaq has announced
http://www.openvms.digital.com/OPENVMS/strategy.html a 5-year plan of
developments in OpenVMS. Of course, Compaq management have yet to prove
that they understand all the business issues around selling and supporting
mission critical, 24x365 computing systems and software. I would not write
VMS off yet and I certainly have seen no proof that NT is stable and
functional enough to replace all these VMS systems.

UNIX
Now the hot UNIX is Linux. UNIX has a strong hold in many application areas
and I am sure will be around for quite a while, at least as long as VMS.
With Linux it is looking as though the UNIX community is learning how to
develop and support a single UNIX instead of all the different varieties
that have been out there. At last UNIX could compete in breadth of
applications and users with Windows, but that would take many years. Also,
in the last resort, you can adapt, fix or whatever the kernel. not possible
with the others.

Windows NT
Windows and Windows NT has and will certainly have for many years the
majority of the computers. However, the majority of the computers in this
world are not in the factory. They are in the home and office. It is from
the these desktops that the e-commerce that excites the investors so much
will come. The plant floor operator will not be ordering on the web between
batches of product! Why then should Windows development pay attention to
our needs first, if at all?

When the mission critical functions are in PLCs then the HMI can be more
freely chosen. When you need to put code in the loop that is not available
in the PLC, then one needs a computer in the loop and you then have a
mission critical computer system.

There is no one operating system that will meet all needs, there never has
been and there most likely never will be. Variety and choice are essential
if we are to improve the quality and productivity of the product and plant
and this means not doing the same old thing that everyone else continues to
do. Real innovation is the key.

Peter Clout

Peter Clout
Vista Control Systems, Inc.
176 Central Park Square
Los Alamos, NM 87544-3012
(505) 662-2484
FAX (505) 662-3956
[email protected]
http://www.vista-control.com

D

#### David Johns

As a corollary:

I have only recently starting using Linux and I haven't found it to be
either stable or easy to install.

To start, I doubt I would have been able to install it without a very useful
guide written by a reputable computer magazine. I still haven't been able
to get X-Windows to display anything other than 640x480 16 colour screen and
it took me three hours to figure out which text file to edit and which
program to run in order to get 'Lilo' to boot Win98 as the default OS. A
number of times, while looking through the system to find the previously
mentioned text file, the operator interface locked up (producing numerous
core dumps) and required a complete system restart. Once I had to turn the
PC off without shutting down and the subsquent (& automatic) build of the,
now corrupted, file system took nearly half an hour.

This was all with the latest version of one of the most popuplar
distrubitions and its default user interface.

My initial reaction, however, was not 'Linux is a terrible OS' and to
to go out and buy some books and invest some time in learning about it.

Alternatively, I have never found Windows or WinNT to be anything other than
stable, cheap and easy-to-use operating systems. With Windows, I can put
the CD in, start the PC and be up an running with full access to the systems
capabilities, in 20 minutes. It's all a matter of familiarity.

My major concern with the Microsoft 'monopoly' is that the environment it
has created had damaged peoples critical-thinking abilities. Too often I
have heard people blame Windows/Microsoft/Intel for problems which
invariably exist with the packages that they are running. Recently I had an
experience where a networking expert had us chasing our tails looking for
problems in completely the wrong place. His preference for Unix over NT
(and his lack of understanding NT) made him blame Microsoft first and think
later.

What is worse, is that I know software engineers who have written software,
found it doesn't work, decide its a Microsoft problem and leave the bugs in
their program. They then explain to the customer that their program doesn't
work because Windows/WinNT is a poor product. Quite a number of customers
have acutally agreed with this assessment and are quite happy to also blame
Microsoft for their wasted money and lack of a working product.

Regards

David Johns
Process Control Engineer

B

#### Blunier, Mark

> I have only recently starting using Linux and I haven't found it to be
> either stable or easy to install.
>
I don't want to start another flame war, so I'll try to keep this
short.

If you want a distribution that is easy to install, I suggest you
take a look at the Corel distribution. One of the guys did a demo
install of it at our last users group meeting, and I was amazed at
how easy it installed. If I remember correctly, he had to answer two
questions to the effect of 1) should Linux take over the entire
hard drive? He answered yes. 2) what kind of install - basic, full,
complete, and he had a system that had detected his sound card, network
card, video card, and came up in a 1024x768x16 X window.

Mark Blunier
The opinions expressed in this message are not necessarily those of the
company.

M

#### Mark Hutton

To put my earlier comment into context, I am NOT a Unix/Linux head.

My experience of Unix which amounted to 18 mths 50/50 with Windows 3.11 was
of HP-UX and was not a very good experience. It seemed to crash as often as
Windows and took about half an hour or more to reboot. (Linux reboots as
quickly as Windows 98).

I work daily (6 to 15 hours a day) with Windows 95/98, and part of my job is
to provide system support in the office where I work - though I don't
conisder myself a windows guru. I also build my own PCs at home and upgrade
the hardware every three or four months.

I have a total of about 20 hours experience with Linux, including installing
it three times.

I have never had to reboot the (Linux) system without shutting it down,
CTRL-BACKSPACE has allways worked. (Also, I never had the problems many be
people reported installing OS/2).

Windows 98 has crashed for whatever reason, atleast three times a week,
though some times five or more times a day. As I have said in other threads
though this is probably not entirely the fault of windows, though a
correctly designed operating system would not crash because of badly
designed programs.

Regards
Mark

D

#### Dave Ferguson

I do not want this to come of as flame because that is not the intention....this is a pet peave and misconception of many people and I feel
needs a response..............

I had to reply to this one. I have been a network admin full time for roughly 5 years. I have an MCSE but by no means a "Microsoft weanie" but I
laugh when people say that "it crashes up to 5 times a day". What are you running, I have found that the vast majority (95% or better) of the problems with Windows are with people who don't understand how it works or try to install software that is shaky or answer instalation questions that they don't understand etc..

A perfectly good example of this is we had systems that crashed 1-2 times a month in production control areas of our plant and we implemented system policies in Windows (Policy Editor)which locked down access to the ability
to "tweak" windows. We have yet to have a crash on these systems since.

A colleague of mine is in a NT 4.0 class right now and just told me about being told you could not lock down windows 98, bull**, we have some machines so tight all you can do is run 1 program or reboot, read the resource kit

While I agree that a good Operating System should function and for all practical purposes NT and 98 I run 24-7-365 and Operate sections of our
plant that are responsible for $50 million in raw materials and have had little to no trouble other than failed hardware (twice in 2 years, HD and monitor). As in most things in life if you know what you are doing, you have little trouble. Do I consider myself an expert, no, I just read the books and put the time in. Everyone wants to take the easy way out and run with no time, and everyone is a closet expert because they have e-mail at home. Dave Ferguson Blandin Paper Company UPM-Kymmene J #### Jim Redman What I find interesting about this approach,and the general responses, is that they all seem to rely on closely controlling the conditions and limiting the machines. As any good control engineer will tell you, almost any system is stable in steady-state, the real test of the robustness of a system is how it behaves under perturbation. Is it fair to say that Windows is stable when disturbed? Jim -- Jim Redman (505) 662 5156 http://www.ergotech.com R #### R A Peterson << I do not want this to come of as flame because that is not the intention....this is a pet peave and misconception of many people and I feel needs a response..............>> You can consider my response a flame in your direction if you want. << I had to reply to this one. I have been a network admin full time for roughly 5 years. I have an MCSE but by no means a "Microsoft weanie" but I laugh when people say that "it crashes up to 5 times a day". What are you running, I have found that the vast majority (95% or better) of the problems with Windows are with people who don't understand how it works or try to install software that is shaky or answer instalation questions that they don't understand etc..>> Let me give you a few examples of NT problems and the solutions that eventually fixed them i have run across in the last few years. First example. A Compaq computer with NT factory loaded, in a MS approved configuration. This system refused to run for more then a few hours. After much aggravation, and no help from Compaq or MS, the tech support people at Wonderware were kind enough to tell us the the SCSI card on that particular machine was known to not work under NT for some reason, despite being certified, no doubt by professionals. A replacement card fixed it. No configuration problems, no incompetents loading shaky software and entering bad information. Second example. A Micron system that ran fine for several years then began to get slower and slower. Eventually it got to the point where it would take 5 seconds or more to switch tasks. This was tolerable but the BSOD that occurred about once a month wasn't. eventually. a Micron tech suggested the disk was likely badly fragmented and that this could cause this exact problem. Wow - a problem MS denied could ever happen on NT. No doubt it never happens to a professional like you. Installing and running Diskeeper fixed that problem. Hasn't happened since. << A perfectly good example of this is we had systems that crashed 1-2 times a month in production control areas of our plant and we implemented system policies in Windows (Policy Editor)which locked down access to the ability to "tweak" windows. We have yet to have a crash on these systems since. A colegue of mine is in a NT 4.0 class right now and just told me about being told you could not lock down windows 98, bull**, we have some machines so tight all you can do is run 1 program or reboot, read the resource kit >> Why bother with a PC when a terminal would do the job? Why have a general purpose computer available if you can't use it for anything else? Thats the whole purpose of having such a beast. << While I agree that a good Operating System should function and for all practical purposes NT and 98 I run 24-7-365 and Operate sections of our plant that are responsible for$50 million in raw materials and have had little to no trouble other than failed hardware (twice in 2 years, HD and monitor).

As in most things in life if you know what you are doing, you have little trouble. Do I consider myself an expert, no, I just read the books and put the time in. Everyone wants to take the easy way out and run with no time, and everyone is a closet expert because they have e-mail at home.

Dave Ferguson
Blandin Paper Company
UPM-Kymmene
>>

Bob Peterson
Computer Amateur with Email at Home

J

#### Johan Bengtsson

I would say that it heavily depends on what you mean by "disturbed".

I think it is easier to make a disturbance that will make windows (especially the 9x versions, to a lesser degree NT) that would be a real disturbance to the control part than it would be to make a disturbance to a PLC causing the same disturbance to the control.

What do I say with this? A windows NT system working fine (ie correct drivers for the devices, no hardware errors, no interrupt collisions
and so on) is probably quite stable and can be running for a really long time. A small disturbance from a somewhat misbehaving program
is probably not affecting the system that much. (it may affect determinism issues some, but not nesecarily too much). A small hardware problem (like for example a lost network cable) can make the hardware driver to that device disturb the system quite heavily.

Another example (from my computer here at work). Every time I insert a CD, or the system thinks I may be trying to access the CD (like when
changing directory and NT wants to know the name of the CD on order to display it). Sometimes almost everything in my computer stops for about
4-5 seconds while the driver spins up the CD and reads the needed information. I have understood this as a driver issue (ATAPI) (I don't really bother enough to do something about it, but if the computer would have been controlling something then...)

It probably gets bader in windows 9x, probably not because the operating system by itself, but becase a somewhat misbehaving application or driver can affect the entire system more. For example a system or GDI resource leak make the computer slower over time until it is restarted and isn't something you will generally notice until you are there...

In a PLC such disturbances is hard(er) to introduce, you don't have a CD if you have a network card it probably don't disturb the PLC in the same way and so on.

I would say that a PC is less stable given the possible (easily introduced) disturbances than a PLC. Another point is that the more code in the OS and control engine, the higher probability for bugs. NT may be tested more than a PLC, but it is also very much bigger than the controlling code in the PLC.

A PC based system can be used, but is perhaps not good for really critical applications. And if you plan to really use the fact that it is a PC and
let people do other things than just controlling you introduce a significantly higher probability for things to go wrong.

/Johan Bengtsson

----------------------------------------
Box 252, S-281 23 H{ssleholm SWEDEN
Tel: +46 451 49 460, Fax: +46 451 89 833
E-mail: [email protected]
Internet: http://www.pol.se/
----------------------------------------

M

#### moores3

This is not an example of a system only being stable in a "steady state".
The issue is one of keeping people from corrupting the performance of the
system. If you leave a programming PC connected to a PLC network and allow
the wrong people to make changes then you will have unplanned crashes. Does
this mean that the PLC is only good in a steady state?

If you allow users to log in to a UNIX system as "root" and you leave them
access to change files in /etc or other critical directories, then you will
have crashes.

B

#### Bill Code

[Bill Code] I use NT at home and at the office. I use it as a development platform and as a target platform for delivery of custom industrial software. I have supplied many automation systems hosted on it. It works 24/7 reliably.

Let me give you a few examples of NT problems and the solutions that eventually fixed them i have run across in the last few years.

First example. A Compaq computer with NT factory loaded, in a MS approved configuration. This system refused to run for more then a few hours. After much aggravation, and no help from Compaq or MS, the tech support people at Wonderware were kind enough to tell us the the SCSI card on that particular machine was known to not work under NT for some reason, despite being certified, no doubt by professionals. A replacement card fixed it. No configuration problems, no incompetents loading shaky software and entering bad information.
[Bill Code] Are you sure you can call this a problem with NT or was it a problem with somebody's SCSI card?

Second example. A Micorn system that ran fine for several years then began to get slower and slower. Eventually it got to the point where it would take 5 seconds or more to switch tasks. This was tolerable but the BSOD that occurred about once a month wasn't. eventually. a Micron tech suggested the disk was likely badly fragmented and that this could cause this exact
problem. Wow - a problem MS denied could ever happen on NT. No doubt it never happens to a professional like you. Installing and running Diskeeper fixed that problem. Hasn't happened since.
[Bill Code] AFAIK disk fragmentation is a maintenance problem with all computer operating systems.

Why bother with a PC when a terminal would do the job? Why have a general purpose computer available if you can't use it for anything else? Thats the whole purpose of having such a beast.
[Bill Code] My company uses PCs because you can purchase a great deal of computing power (in terms of throughput and storage) for a reasonably
low cost. With the addition of well developed and well tested custom application software, the result is low cost, fit for purpose industrial systems.

------- End of forwarded message -------

Bill Code
MPM Engineering Ltd.
4-6240 202nd St., Langley, B.C., Canada, V2Y-1N2
Phone: 604-534-6605 Fax: 604-534-6693
E-Mail: [email protected] WWW: http://WWW.MPM-ENG.COM

M

#### Mark Blunier

<[email protected]> wrote:
> This is not an example of a system only being stable in a "steady state". The issue is one of keeping people from corrupting the performance of the system. If you leave a programming PC connected to a PLC network and allow the wrong people to make changes then you will have unplanned crashes. Does this mean that the PLC is only good in a steady state?<

No, it means someone (not the vendor) messed up. I leave the programmer connected to the pc all the time, but as only the people that are trained to use it, use it, it isn't a problem.

> If you allow users to log in to a UNIX system as "root" and you leave them access to change files in /etc or other critical directories, then you will have crashes.<

No kidding. But then, the MMIs that I've seen running on UNIX systems do not give root access to users.

Releasing MMI software that has, as the default setup, open access to stuff that can break the system is incompetence by the MMI vendor, and
yet, this is way many MMIs default, and encourage people to try to set things up to there own like, fix things they think are 'broke', and etc.
Engineers shouldn't be asking the question what can I do to keep the operators from using explorer, or the 'my computer' icon, they should
have to ask, what can I do to give access to 'my computer' back to the operator. As I see it, common sense dictates if the operators don't need
it, it could break the system, and people are inclined to play with it, then the operators shouldn't be given the open invitation to play it.

Mark Blunier
Any opinions expressed in this message are not necessarily those of the
company.

R

#### R A Peterson

[email protected] writes:
<< ...then nothing unusual would happen. If the controlling part of software is properly designed and process priorities are properly
assigned, inserting a CD in a tray and spinning it up definitely cannot bring any significant delays to the control programs. >>

Maybe in a properly designed OS, but certainly not Win95 or 98. I have a laptop at work that when I insert a CDROM, the mouse does its little dance for at least ten seconds before I can do anything else. I doubt that a control program would work correctly during that time.

I have seen similar things on WinNt systesm, where the system basically stops when a disk is removed from a FDD during a read or write to it.

Not inherently a problem of disks, but of the OS.

M

#### moores3

With any system you have to install your software and make sure that it works. I don't see any difference in UNIX or NT. I don't know why people keep using examples of their PC on their desk to explain problems with NT. Like I said before if you give the end user root privs and let them have at a UNIX system you will get a crash. I have had product NT boxes run for ever without crashing. I have had the same with UNIX.

Jim Redman <[email protected]> on 12/17/99 03:21:09 PM wrote:
>>
From: [email protected]
> This is not an example of a system only being stable in a "steady state". The issue is one of keeping people from corrupting the performance of the system.<

It seems that the issue is not performance per se, but simple stability. ...<<

J

#### Jim Redman

From: [email protected]
> This is not an example of a system only being stable in a "steady state". The issue is one of keeping people from corrupting the performance of the system.<

It seems that the issue is not performance per se, but simple stability.

It may make sense to limit access to the machine to stop the operator playing "Doom" but that's not the issue. What I'm hearing is that it's
very important to control what is installed, when and how with Microsoft systems to avoid a crash.

While such control and limits do allow 24/7 execution it is not ideal for a production system. As a simple example, suppose that one morning you discover a virus. This virus will systematically crash the systems unless you install anit-virus software, and yet it is not totally clear that installing the anti-virus software will not itself cause the system
to crash.

There is also a "lost opportunity cost" since people may be unwilling to install enhancements or add-on products that would help production
fearing that these may cause the system to crash.

Each is an example of a lack of robustness. I don't know if any system is currently robust enough to withstand changes of this type, but as
users we should not complacently accept the situation and quietly find what are essentially work arounds but should yell loud and long that
this is not acceptable and that the quality and robustness should be improved. In fact, I think that in Redmond this message is being received, which, if they can execute, is good for everyone.

Jim
--
Jim Redman
(505) 662 5156
http://www.ergotech.com

V