QNX 6 driver support is NOT BETTER than QNX 4

When we were in QNX 4, we were facing with problems of installing the O/S
into a new hardware, especially server class hardware. Our solution was to
buy server machine which has integrated SCSI and LAN adaptor, and install an
extra OLD SCSI and LAN cards. Thus our customer always asked us “Are we
buying an already terminated O/S?”.

We thought this problem would be solved if we have migrated our software to
QNX 6. But today we encounter the same problem. We have HP Proliant ML350
with integrated SCSI and LAN adaptor, and it doesn’t work. QNX 6.2.1A has
not supported these adaptors. And we are sick of getting the same question
from our customers, so we pass the question to QNX: “ARE WE BUYING A
TERMINATING O/S?”

QNX is not a server-class OS. Version 4 was not, and version 6 is not. It
would be naive to expect that simply going to newer version would change
that. If there are good reasons for you to use an OS that was designed for
something else in your server-class system, those reasons better be good
enough to justify getting your hardware supported through custom engineering
channel.

I would like nothing more than see a better situation with hardware support
by QNX myself. However one has to face the reality. It is unrealistic to
expect adequate hardware support from a company with ~100 employees for a
platform (x86) that generates the least portion of their revenues. Every
time somebody asks this question I bet someone at QNX gets an itch to remove
self-hosted x86 platform altogether.

Much as I don’t like to say that, you’ll be better off with Linux or Solaris
(Sun has renewed interest in Solaris on x86), depending on your needs. It
will be slower. But then again, an 18-wheeler is slower than a roadster. So
what?

– igor

“Johannes” <Jsukamtoh@infolink.co.id> wrote in message
news:bmfhp8$fnb$1@inn.qnx.com

When we were in QNX 4, we were facing with problems of installing the O/S
into a new hardware, especially server class hardware. Our solution was to
buy server machine which has integrated SCSI and LAN adaptor, and install
an
extra OLD SCSI and LAN cards. Thus our customer always asked us “Are we
buying an already terminated O/S?”.

We thought this problem would be solved if we have migrated our software
to
QNX 6. But today we encounter the same problem. We have HP Proliant ML350
with integrated SCSI and LAN adaptor, and it doesn’t work. QNX 6.2.1A has
not supported these adaptors. And we are sick of getting the same question
from our customers, so we pass the question to QNX: “ARE WE BUYING A
TERMINATING O/S?”

Igor Kovalenko <kovalenko@attbi.com> wrote:

IK > I would like nothing more than see a better situation with hardware support
IK > by QNX myself. However one has to face the reality. It is unrealistic to
IK > expect adequate hardware support from a company with ~100 employees for a
IK > platform (x86) that generates the least portion of their revenues. Every
IK > time somebody asks this question I bet someone at QNX gets an itch to remove
IK > self-hosted x86 platform altogether.

I don’t believe that X86 generates the “least” revenue for QSSL. They,
and you, make it sound like anyone still using X86 is a dinosour. I
believe that many/most people are still using X86.

That aside, I too wish there were better hardwre support from QSSL. I
don’t blame QSSL. They do a good job for the number of people that
they do have. But it’s not just people alone. I have this new scremer
of a system, a Dell. I have some problems and QSSL got me past most of
them. Thanks, Hugh and Barry F.! But it also has this Nvidia video
card. I’m told it’s a state of the are card. But Nvidia won’t share
the art. There are no specs for QSSL to write a video driver for it.

So, when someone is looking for a good hardware/software fit, they just
need to take all of this into account and keep shopping until they find
just the right fit.

With this in mind I have one final comment to anyone else looking for
their next peice of hardware. Many of the big name vendors are more
likely to have proprietary hardware and undocumented software
requirements. They develope their own techniques for doing everything
because they can often do it just a little bit faster that way. The
lesser brands and brand-X vendors usually stay pretty close to a
standard if they can. Except for the real cheap ones that just skimp
on everyhing. Broadbased examples: WinModems and now WinPrinters.

Bottom line, whatever you buy, make sure you can return it.

Good Luck & Don’t Blame QNX

Bill Caroselli wrote:

Igor Kovalenko <> kovalenko@attbi.com> > wrote:

IK > I would like nothing more than see a better situation with hardware support
IK > by QNX myself. However one has to face the reality. It is unrealistic to
IK > expect adequate hardware support from a company with ~100 employees for a
IK > platform (x86) that generates the least portion of their revenues. Every
IK > time somebody asks this question I bet someone at QNX gets an itch to remove
IK > self-hosted x86 platform altogether.

With this in mind I have one final comment to anyone else looking for
their next peice of hardware. Many of the big name vendors are more
likely to have proprietary hardware and undocumented software
requirements. They develope their own techniques for doing everything
because they can often do it just a little bit faster that way. The
lesser brands and brand-X vendors usually stay pretty close to a
standard if they can. Except for the real cheap ones that just skimp
on everyhing. Broadbased examples: WinModems and now WinPrinters.

Bottom line, whatever you buy, make sure you can return it.

Good Luck & Don’t Blame QNX

Personally, I don’t get my knickers in a twist at all with hardware support
(even on QNX4). The point is, if QNX brings something to the table that
makes your product better, then you make the hardware fit the O/S (not the
other way around). It isn’t hard to do, it’s not like there is a shortage
of hardware vendors out there.

“Rennie Allen” <rallen@csical.com> wrote in message
news:bmh2tu$jcj$1@inn.qnx.com

Bill Caroselli wrote:
Igor Kovalenko <> kovalenko@attbi.com> > wrote:

IK > I would like nothing more than see a better situation with hardware
support
IK > by QNX myself. However one has to face the reality. It is
unrealistic to
IK > expect adequate hardware support from a company with ~100 employees
for a
IK > platform (x86) that generates the least portion of their revenues.
Every
IK > time somebody asks this question I bet someone at QNX gets an itch
to remove
IK > self-hosted x86 platform altogether.

With this in mind I have one final comment to anyone else looking for
their next peice of hardware. Many of the big name vendors are more
likely to have proprietary hardware and undocumented software
requirements. They develope their own techniques for doing everything
because they can often do it just a little bit faster that way. The
lesser brands and brand-X vendors usually stay pretty close to a
standard if they can. Except for the real cheap ones that just skimp
on everyhing. Broadbased examples: WinModems and now WinPrinters.

Bottom line, whatever you buy, make sure you can return it.

Good Luck & Don’t Blame QNX

Personally, I don’t get my knickers in a twist at all with hardware
support
(even on QNX4). The point is, if QNX brings something to the table that
makes your product better, then you make the hardware fit the O/S (not the
other way around). It isn’t hard to do, it’s not like there is a shortage
of hardware vendors out there.

Have you found any vendors with tape drives that are supported Rennie?
Any SCSI RAID controllers?

It is generally hard to do ‘make the hardware fit the OS’ approach from
business/management perspective. In case you have forgot, hardware is harder
to change, because it is HARDware. When a project starts and deadline is
there, nobody has time to mess around trying to find the hardware that ‘fits
the OS’. Especially when ‘fitting’ hardware tends to be obsolete/EOL…

– igor

“Bill Caroselli” <qtps@earthlink.net> wrote in message
news:bmgpf3$d2g$1@inn.qnx.com

Igor Kovalenko <> kovalenko@attbi.com> > wrote:

IK > I would like nothing more than see a better situation with hardware
support
IK > by QNX myself. However one has to face the reality. It is unrealistic
to
IK > expect adequate hardware support from a company with ~100 employees
for a
IK > platform (x86) that generates the least portion of their revenues.
Every
IK > time somebody asks this question I bet someone at QNX gets an itch to
remove
IK > self-hosted x86 platform altogether.

I don’t believe that X86 generates the “least” revenue for QSSL. They,
and you, make it sound like anyone still using X86 is a dinosour. I
believe that many/most people are still using X86.

With QNX4, yes.

With QNX6, I don’t doubt at all that it is the last platform in terms of
revenues. They just took too long to bring it up. By the time QNX6 reached
the state it could replace QNX4, everyone who wanted to replace QNX4 in the
first place, has already done so - with Winows or Linux. Non-x86 projects
are different story - there was no QNX4 for them :wink:

With this in mind I have one final comment to anyone else looking for
their next peice of hardware. Many of the big name vendors are more
likely to have proprietary hardware and undocumented software
requirements. They develope their own techniques for doing everything
because they can often do it just a little bit faster that way. The
lesser brands and brand-X vendors usually stay pretty close to a
standard if they can. Except for the real cheap ones that just skimp
on everyhing. Broadbased examples: WinModems and now WinPrinters.

What you’re missing is that nobody cares whether a vendor follows
‘standards’ or not, so long as they provide software support. If you stay
with ‘mainstream’ OS, be that Windows or Linux, the question of drivers for
COTS hardware just does not exist. When you bring QNX onto the table, the
question usually pops up and NOBODY likes to have to deal with that problem.

I have to say, if your business case is reasonable and the hardware is
popular enough among their potential customers and specs are available,
QNX likely will provide a driver. They can even do it at no charge if
situation warrants… The trouble is, even that case is still more legwork
than most project managers want to have. Negotiations, legalese, etc… and
after all you’re getting a driver that is not tested by a lot of people -
which means you might have problems with it later. This can only work out if
a customer has some insider advocat like one of us here, who can oil the
hinges when they squeak and push the carriage when it is stuck…

So, whichever way you look at it, the picture complicated. QNX really works
out better for projects that involve custom hardware or
new-just-out-of-design hardware that has no drivers for any OS yet anyway…
I believe QNX has been concentrating on that kind of projects lately and I
don’t blame them.

– igor

Igor Kovalenko wrote:

Have you found any vendors with tape drives that are supported Rennie?
Any SCSI RAID controllers?

Your point that QNX is not a server OS is well taken. I wasn’t trying
to argue that it is a server OS.

hardware is harder

to change, because it is HARDware.

Not in the real world where hardware vendors are a dime a dozen. There
are basically 2 markets: custom designed boards, where the customer
literally does design the hardware for the software, and semi-custom
embedded where the customer buys the hardware to fit the software. I
have done both, and never encountered a serious issue with getting
supported hardware.

… When a project starts and deadline is

there, nobody has time to mess around trying to find the hardware that ‘fits
the OS’. Especially when ‘fitting’ hardware tends to be obsolete/EOL…

Yes I am aware that there are badly managed projects out there :slight_smile: but I
don’t think that the point of the discussion was to advocate for poor
management.

Would I like to see every piece of hardware supported under QNX ? sure why
not ? Would I like a Gulfstream IV ? sure.

If I browse this forum, over 50% of the people here are x86 user if I am not
wrong. That could mean either we have more x86 user here or we have more
problems with x86! Well, in either case, QSSL has to spend a bit more time
in x86.

Switch to Solaris? That idea has been in my mind before I migrate to
Neutrino. And looks like I have no choice now. I don’t want to repeat the
same problem that I have in the past few years looking for a computer part
that suit the O/S, especially in my area which is hard to find old stuffs.
It will be nice if we have a computer junk yard here, like the way I do with
my old car. Talking about car, will you buy an older release car just
because you don’t know how to drive the new version? Or will you learn to
live with the new one? Hi-tech people will not accept this! We have new
release of computer in every quarter of a year, which also mean an older
version is obsolete in every quarter year as well. So if you can afford to
buy a new one, why should you buy the older one? If others can do it, why
can’t QSSL?

We have been a good fan of QNX (since 1995). And we hope QSSL will keep up
to date with new hardware releases. Not all but at least one big brand like
HP or Dell. We are talking about highly availability and highly reliability
software here running 7x24, so none of our customers will dare to use
something that they have never heard of (sorry to other compatibles). That
is too risky. Even for big brand, our customer dare not use the desktop.
They would rather go for server class. BTW they also ask why QNX? They have
never heard of it until we introduced QNX to them, and that took a lot of
our effort.

“Rennie Allen” <rallen@csical.com> wrote in message
news:bmh2tu$jcj$1@inn.qnx.com

Bill Caroselli wrote:
Igor Kovalenko <> kovalenko@attbi.com> > wrote:

IK > I would like nothing more than see a better situation with hardware
support
IK > by QNX myself. However one has to face the reality. It is
unrealistic to
IK > expect adequate hardware support from a company with ~100 employees
for a
IK > platform (x86) that generates the least portion of their revenues.
Every
IK > time somebody asks this question I bet someone at QNX gets an itch
to remove
IK > self-hosted x86 platform altogether.

With this in mind I have one final comment to anyone else looking for
their next peice of hardware. Many of the big name vendors are more
likely to have proprietary hardware and undocumented software
requirements. They develope their own techniques for doing everything
because they can often do it just a little bit faster that way. The
lesser brands and brand-X vendors usually stay pretty close to a
standard if they can. Except for the real cheap ones that just skimp
on everyhing. Broadbased examples: WinModems and now WinPrinters.

Bottom line, whatever you buy, make sure you can return it.

Good Luck & Don’t Blame QNX

Personally, I don’t get my knickers in a twist at all with hardware
support
(even on QNX4). The point is, if QNX brings something to the table that
makes your product better, then you make the hardware fit the O/S (not the
other way around). It isn’t hard to do, it’s not like there is a shortage
of hardware vendors out there.

“Rennie Allen” <rallen@csical.com> wrote in message
news:bmhamt$oqf$1@inn.qnx.com

Igor Kovalenko wrote:

Have you found any vendors with tape drives that are supported Rennie?
Any SCSI RAID controllers?

Your point that QNX is not a server OS is well taken. I wasn’t trying
to argue that it is a server OS.

hardware is harder
to change, because it is HARDware.

Not in the real world where hardware vendors are a dime a dozen. There
are basically 2 markets: custom designed boards, where the customer
literally does design the hardware for the software, and semi-custom
embedded where the customer buys the hardware to fit the software. I
have done both, and never encountered a serious issue with getting
supported hardware.

I dare you to find half a dozen vendors that provide a high-density (say
16) general purpose DSP pool board in PTMC format (that is, PCI telecom
mezzanine card, PICMG 2.15). When you’re done, find me half dozen cPCI CPU
cards with dual PTMC sites. That does not sound like too specialized
hardware, right?

I am not even asking to find supported ones. Just any.

. When a project starts and deadline is
there, nobody has time to mess around trying to find the hardware that
‘fits
the OS’. Especially when ‘fitting’ hardware tends to be obsolete/EOL…

Yes I am aware that there are badly managed projects out there > :slight_smile: > but I
don’t think that the point of the discussion was to advocate for poor
management.

Poor management? Wake up call Rennie. We’re living in an economic
environment where management choices are driven almost entirely by customer
requests. Nobody wants to do almost any development that does not have a
prospective customer waving a checkbook attached to it…

They have funky terms for it now. ‘Cash oriented culture’, et cetera. For
better or for worse, it appears to separate companies with stock below and
above junk threshold these days. Of course the trouble is, when a customer
with checkbook materializes, he does not want to hear about 2 year long
development. He wants it in 6 months, or you don’t have a deal.

– igor

“Johannes” <Jsukamtoh@infolink.co.id> wrote in message
news:bmiaqb$ff7$1@inn.qnx.com

If I browse this forum, over 50% of the people here are x86 user if I am
not
wrong. That could mean either we have more x86 user here or we have more
problems with x86! Well, in either case, QSSL has to spend a bit more time
in x86.

No, it is just that most people who write here are those that have some
passion for QNX, rather than ‘just a job’ attitude. And that sort of people
tends to run it on their desktops/laptops… not because they really need
to. Becase they can :wink: We also report more problems - not because x86 has
more problems, but because other platforms don’t get [ab]used in the ways we
can do it on x86. This helps to uncover lots of problems that are relevant
for all platforms, we just hardly see them reported.

So yes, looking at the newsgroups one might conclude that majority of the
users are on x86. In reality I think the majority is simply not here,
because they are ‘just a job’ folks.

Switch to Solaris? That idea has been in my mind before I migrate to
Neutrino. And looks like I have no choice now. I don’t want to repeat the
same problem that I have in the past few years looking for a computer part
that suit the O/S, especially in my area which is hard to find old stuffs.
It will be nice if we have a computer junk yard here, like the way I do
with
my old car. Talking about car, will you buy an older release car just
because you don’t know how to drive the new version? Or will you learn to
live with the new one? Hi-tech people will not accept this! We have new
release of computer in every quarter of a year, which also mean an older
version is obsolete in every quarter year as well. So if you can afford to
buy a new one, why should you buy the older one? If others can do it, why
can’t QSSL?

Who are the others? Linux has it partially because lot of ‘dot-com money’
has been funneled into it and partially because there is a lot more people
doing drivers - in many cases for free. Solaris (on x86) does not have the
same level of hardware support as Windows or even Linux.

Part of the problem of course is that driver DDKs for QNX did not exist for
too long, and some still do not exist (block drivers, hello Kevin). When
they appeared, it turned out QNX has a separate DDK for every subsystem and
in some cases they don’t look anything like each other. In Solaris you have
DKI/DDI and STREAMS that work for almost any kind of driver, using the same
set of concepts and functions. They also have white papers describing the
process of porting NT drivers down to tiniest details. QNX has just began
systematizing the eclectic mess they have created on the drivers’ front.

Even if they fix that, I doubt they will be able to catch up with Linux or
even Solaris. QNX lacks recognition among the hardware vendors and they also
lack proper internal attitude wrt making any kind of assurances in the
hardware support area - even simple things like publishing a list of
brands/models that are known to work turns out to be too much to ask. For
some reason Sun is not afraid to publsh a list of laptops that are known to
work with Solaris, even though they know laptops do change often.

We have been a good fan of QNX (since 1995). And we hope QSSL will keep up
to date with new hardware releases. Not all but at least one big brand
like
HP or Dell. We are talking about highly availability and highly
reliability
software here running 7x24, so none of our customers will dare to use
something that they have never heard of (sorry to other compatibles). That
is too risky. Even for big brand, our customer dare not use the desktop.
They would rather go for server class. BTW they also ask why QNX? They
have
never heard of it until we introduced QNX to them, and that took a lot of
our effort.

I believe you, but then you’re counter-arguing yourself. If your customers
don’t dare to use anyting unknown, they should not be using QNX. If they
need a high-availability system, you should not be pushing QNX either
because it is not. Pushing a system that we happen to like personally as a
solution for all problems is not an uncommon pitfall between us here :slight_smile:

Realistically speaking, an OS designed for HA systems needs to support a
whole lot more than just modern hardware (also, contrary the impression QNX
is trying to make, a system does not automatically qualify as a HA just by
virtue of being based on micro-kernel). It needs to support storage
clustering, redundant hot-swappable CPU modules and peripheral cards,
checkpoints, alternate pathing, load balancing, remote management and
whatnot. There is an emerging standard for that, driven by Service
Availability Forum. If you really need that kind of system, you will do
yourself a greatest favor by switching to a platform that was designed for
that by the vendor. Sun has SPARC server boxes that support most of that
right out of box. They will cost you, but you get what you’re paying for.
Neither x86 platform in general, nor QNX are particularly strong in this
area.

– igor

“Rennie Allen” <> rallen@csical.com> > wrote in message
news:bmh2tu$jcj$> 1@inn.qnx.com> …
Bill Caroselli wrote:
Igor Kovalenko <> kovalenko@attbi.com> > wrote:

IK > I would like nothing more than see a better situation with
hardware
support
IK > by QNX myself. However one has to face the reality. It is
unrealistic to
IK > expect adequate hardware support from a company with ~100
employees
for a
IK > platform (x86) that generates the least portion of their
revenues.
Every
IK > time somebody asks this question I bet someone at QNX gets an
itch
to remove
IK > self-hosted x86 platform altogether.

With this in mind I have one final comment to anyone else looking for
their next peice of hardware. Many of the big name vendors are more
likely to have proprietary hardware and undocumented software
requirements. They develope their own techniques for doing everything
because they can often do it just a little bit faster that way. The
lesser brands and brand-X vendors usually stay pretty close to a
standard if they can. Except for the real cheap ones that just skimp
on everyhing. Broadbased examples: WinModems and now WinPrinters.

Bottom line, whatever you buy, make sure you can return it.

Good Luck & Don’t Blame QNX

Personally, I don’t get my knickers in a twist at all with hardware
support
(even on QNX4). The point is, if QNX brings something to the table that
makes your product better, then you make the hardware fit the O/S (not
the
other way around). It isn’t hard to do, it’s not like there is a
shortage
of hardware vendors out there.