Re: [9fans] commercial deployment

Re: [9fans] commercial deployment

[anothy]

//Is there or will there ever be commercial/industrial deployment of plan9?

short answer: yes, sort of.
long answer: that depends, what does that mean to you?

there are a number of people who use Plan 9 in their "day jobs", in industry or
commercial environments, even though those jobs likely have nothing to do
with plan 9. i did that when i admined Unix boxes, for example.

as mentioned a few times here, plan 9's been used in several projects in the
labs, like the VPN brick, that may someday see the light of day (hey, i'd buy
one of those if i could).

i believe somebody outside the labs was using Plan 9 in a hardware mpeg
encoder.  is that still going on? which edition?

and, of cource, the Plan 9 and Inferno development are done using it.

while Plan 9 still tends to think of itself as a research operating system, it's
certainly usable in production/commercial environments, and is quite stable -
perhaps moreso than many OSs that _do_ explicitly consider themselves
production quality.

nicely packaged Plan 9 sets - with the manuals and papers printed and nicely
bound and a CD with the distribution and various add-ons - are sold by Vita
Nuova, commercially.

the current license is much friendlier to commercial use than the 2nd edition
on (that is, it allows it at all). i can't figure out how to compare Plan 9's Open
Source license to the other Open Source licenses (cough*GPL*cough) in terms
of suitability for commercial use without risking a flame war.

there is not, however, industry support for Plan 9 the way there is for other
OSs, either in terms of hardware vendors explicitly supporting Plan 9 drivers
or providing such themselves, or in terms of commercial software vendors
writing for Plan 9. nor can you get Plan 9 pre-installed on a new PC (AFAIK).
-α.

Re: [9fans] commercial deployment

[Mike Haertel]

an acquaintance of mine from the university now works for a company
that is using Plan 9 to do high performance media servers (some video
on demand kind of thing, i think).  they were using the 1995 edition,
which they commercially licensed from bell labs.

they needed a real time OS; they looked at Linux at the time (a few
years ago, not sure exactly when) and decided that fixing all the
problems to suit their needs would be too hard given how much of a
bloated pig Linux had become, and so they decided to start with Plan 9
and fix it up with the real-time features they needed.

i don't know the name of the company or whether they're still in
business.  it is/was located somewhere in in the Portland, OR area.  i
heard all this when i ran into the fellow in a restuarant back in
January or so, so it seems likely they're still around.

Re: [9fans] commercial deployment

[Fariborz 'Skip' Tavakkolian]

nCUBE

At 01:36 PM 7/19/2001 -0700, Mike Haertel wrote:
>an acquaintance of mine from the university now works for a company
>that is using Plan 9 to do high performance media servers (some video
>on demand kind of thing, i think).  they were using the 1995 edition,
>which they commercially licensed from bell labs.
>
>they needed a real time OS; they looked at Linux at the time (a few
>years ago, not sure exactly when) and decided that fixing all the
>problems to suit their needs would be too hard given how much of a
>bloated pig Linux had become, and so they decided to start with Plan 9
>and fix it up with the real-time features they needed.
>
>i don't know the name of the company or whether they're still in
>business.  it is/was located somewhere in in the Portland, OR area.  i
>heard all this when i ran into the fellow in a restuarant back in
>January or so, so it seems likely they're still around.
>
>

Re: [9fans] commercial deployment

[Alberto Nava]

Mike Haertel wrote:

> an acquaintance of mine from the university now works for a company
> that is using Plan 9 to do high performance media servers (some video
> on demand kind of thing, i think).  they were using the 1995 edition,
> which they commercially licensed from bell labs.
>
> they needed a real time OS; they looked at Linux at the time (a few
> years ago, not sure exactly when) and decided that fixing all the
> problems to suit their needs would be too hard given how much of a
> bloated pig Linux had become, and so they decided to start with Plan 9
> and fix it up with the real-time features they needed.
>
> i don't know the name of the company or whether they're still in
> business.  it is/was located somewhere in in the Portland, OR area.  i
> heard all this when i ran into the fellow in a restuarant back in
> January or so, so it seems likely they're still around.

The company is called ncube (remember hypecubes...),
http://www.ncube.com

We've done quite a lot of work on Plan9 to get it to work pretty
nicely as a video server OS: copy-less IO, distributed kernel resident
volume manager
and file system, new IP stack, async read and writes, real-time
scheduling,etc.

I'm not sure if we have position open right now, but if you're interested
please check
with pkiddle@ncube.com.

Re: [9fans] commercial deployment

[Ozan Yigit]

mike@ducky.net (Mike Haertel) writes:

> they needed a real time OS; they looked at Linux at the time (a few
> years ago, not sure exactly when) and decided that fixing all the
> problems to suit their needs would be too hard given how much of a
> bloated pig Linux had become, and so they decided to start with Plan 9
> and fix it up with the real-time features they needed.

well, to be fair, there exists a number of "real-time" extensions to linux
with variable levels of usefullness. i looked at about four of them [one
of them with a patent and a ton of apologists :/] and i figure none of them
will make it to OS proper (bazaar? ahahahaha). also none of them, so far as
i can see would satisfy "hard" real time needs which was the reason for my
looksee. (eg. pre-run-time scheduling with constraints) so i can see why it
may be easier to start from scratch, and why nasa uses vxworks for its
missions, and not one of the sucky linux haxtensions.

few months ago, i privately inquired about any real-time work with p9, but
noone in the labs knew of any efforts as such. i would be interested to know
 more...

oz
--
www.cs.yorku.ca/~oz	 | if you couldn't find any weirdness, maybe
york u. computer science | we'll just have to make some!   -- hobbes

Re: [9fans] commercial deployment

[Rick Hohensee]

>
> well, to be fair, there exists a number of "real-time" extensions to linux
> with variable levels of usefullness. i looked at about four of them [one
> of them with a patent and a ton of apologists :/] and i figure none of them
> will make it to OS proper (bazaar? ahahahaha). also none of them, so far as
> i can see would satisfy "hard" real time needs which was the reason for my
> looksee. (eg. pre-run-time scheduling with constraints) so i can see why it

What are "pre-run-time scheduling with constraints" and how does rtlinux
fail there? It sounds like it's pretty hard realtime to me. I was looking
at the IRQ stuff in Linux 1.2.13 though, and I don't see why there just
isn't a _RT_<num>interrupt .

Rick Hohensee
			www.clienux.com

Re: [9fans] commercial deployment

[Boyd Roberts]

From: "Ozan Yigit" <oz@tiger.cs.yorku.ca>
> well, to be fair, there exists a number of "real-time" extensions to linux
> with variable levels of usefullness.

to be fair, the extensions (and the core implementation) make lunix
make it a huge, weeping pile of pus.  real time?  they have so much
crud that only real fast cpu's or real slow real time may work.

Re: [9fans] commercial deployment

[Ozan Yigit]

humbubba@smarty.smart.net (Rick Hohensee) writes:

> What are "pre-run-time scheduling with constraints" and how does rtlinux
> fail there?

this is not the right forum to discuss rtlinux, but we know that what
they call "hard" is not quite hard enough. :) as for pre-run-time scheduling
[which is really the only practical way to provide the predictibility needed
for serious hard rt] the two seminal papers are:

Jia Xu, David Lorge Parnas, Pre-Run-Time Scheduling of Processes with
Exclusion Relations on Nested or Overlapping Critical Sections.

and

Jia Xu, David Lorge Parnas, Scheduling Processes with Release Times,
Deadlines, Precedence and Exclusion Relations", IEEE Transactions on
Software Engineering, vol. 16, no. 3, March 1990, pp. 360-369.

also see, by way of a faq :)

Jia Xu, David Lorge Parnas, Priority Scheduling Versus Pre-Run-Time
Scheduling. 7-23, Real Time Systems, Vol. 18, Jan 2000.

enjoy. oz
--
www.cs.yorku.ca/~oz	 | if you couldn't find any weirdness, maybe
york u. computer science | we'll just have to make some!   -- hobbes

Re: [9fans] commercial deployment

[Rick Hohensee]

>
> humbubba@smarty.smart.net (Rick Hohensee) writes:
>
> > What are "pre-run-time scheduling with constraints" and how does rtlinux
> > fail there?
>
> this is not the right forum to discuss rtlinux, but we know that what
> they call "hard" is not quite hard enough. :) as for pre-run-time scheduling
> [which is really the only practical way to provide the predictibility needed
> for serious hard rt] the two seminal papers are:
>
> Jia Xu, David Lorge Parnas, Pre-Run-Time Scheduling of Processes with
> Exclusion Relations on Nested or Overlapping Critical Sections.
>
> and
>
> Jia Xu, David Lorge Parnas, Scheduling Processes with Release Times,
> Deadlines, Precedence and Exclusion Relations", IEEE Transactions on
> Software Engineering, vol. 16, no. 3, March 1990, pp. 360-369.
>
> also see, by way of a faq :)
>
> Jia Xu, David Lorge Parnas, Priority Scheduling Versus Pre-Run-Time
> Scheduling. 7-23, Real Time Systems, Vol. 18, Jan 2000.
>
> enjoy. oz

Hmmmmmmmmmmmmmmm. Sounds academic.

Rick Hohensee
							www.clienux.com

> --
> www.cs.yorku.ca/~oz	 | if you couldn't find any weirdness, maybe
> york u. computer science | we'll just have to make some!   -- hobbes
>

Re: [9fans] commercial deployment

[Boyd Roberts]

From: "Rick Hohensee" <humbubba@smarty.smart.net>
> Hmmmmmmmmmmmmmmm. Sounds academic.
>

shock, horror!

church/turing thesis anyone?

[9fans] more lucent bad news - another 15,000 to 20,000 jobs to go

[Matt]

Lucent to cut another 15,000 to 20,000 jobs

http://www.theregister.co.uk/content/7/20608.html

Re: [9fans] commercial deployment

[Ozan Yigit]

From: "Rick Hohensee" <humbubba@smarty.smart.net>
> Hmmmmmmmmmmmmmmm. Sounds academic.

hmm, right, the same way "priority inheritence protocols" (Sha et al.)
for the priority inversion problem is "academic." :-]

oz

Re: [9fans] commercial deployment

[Ozan Yigit]

rob@plan9.bell-labs.com (rob pike) writes [on priority inheritence
protocols in response to priority inversion problem]

> An academic solution to an academic problem.

evidently the problem arise independently in other circles, well before
it became famous with the pathfinder/vxworks. moreover any problem that
can lose about $430+ million of US taxpayer money is probably no longer
academic. :)

you are right, the whole mess can and should be avoided [eg. my earlier
refs].

oz

Re: [9fans] commercial deployment

[Rick Hohensee]

>
> This is a multi-part message in MIME format.
> --upas-odkgklfjzdotnlpxfzoohjyzvb
> Content-Disposition: inline
> Content-Type: text/plain; charset="US-ASCII"
> Content-Transfer-Encoding: 7bit
>
> > hmm, right, the same way "priority inheritence protocols" (Sha et al.)
> > for the priority inversion problem is "academic." :-]
>
> An academic solution to an academic problem.  Ever notice that Plan 9
> has process priorities that influence scheduling, but interrupts are
> either on (spllo) or off (splhi)?  That simplification, which was a
> step I took when I wrote the very first version of the kernel, fixed a
> problem I didn't even know existed at the time.  Another blow struck
> in the fight for simplicity.
>
> -rob
>

Fight the good fight!

Rick Hohensee

Re: [9fans] commercial deployment

[Boyd Roberts]

> ... but interrupts are either on (spllo) or off (splhi)?

debugging an ipl screwup can take an eternity, so that's
a good case for simplicity.

Re: [9fans] commercial deployment

[rob pike]

[-- Attachment #1: Type: text/plain, Size: 525 bytes --]

> hmm, right, the same way "priority inheritence protocols" (Sha et al.)
> for the priority inversion problem is "academic." :-]

An academic solution to an academic problem.  Ever notice that Plan 9
has process priorities that influence scheduling, but interrupts are
either on (spllo) or off (splhi)?  That simplification, which was a
step I took when I wrote the very first version of the kernel, fixed a
problem I didn't even know existed at the time.  Another blow struck
in the fight for simplicity.

-rob


[-- Attachment #2: Type: message/rfc822, Size: 1609 bytes --]

From: Ozan Yigit <oz@tiger.cs.yorku.ca>
To: 9fans@cse.psu.edu
Subject: Re: [9fans] commercial deployment
Date: Thu, 26 Jul 2001 08:48:35 GMT
Message-ID: <ycdsnfkpxlu.fsf@tiger.cs.yorku.ca>


From: "Rick Hohensee" <humbubba@smarty.smart.net>
> Hmmmmmmmmmmmmmmm. Sounds academic.

hmm, right, the same way "priority inheritence protocols" (Sha et al.)
for the priority inversion problem is "academic." :-]

oz

Re: [9fans] commercial deployment

[bwc]

[-- Attachment #1: Type: text/plain, Size: 255 bytes --]

Plan 9 was the host system while developing the PIX.
If the current license had been available in 1995 there would be
some small parts of Plan 9 code in the PIX.  The code structure
of the PIX is closer to the Oberon system than Plan 9.

  Brantley

[-- Attachment #2: Type: message/rfc822, Size: 2066 bytes --]

From: Steve Kilbane <steve@whitecrow.demon.co.uk>
To: 9fans@cse.psu.edu
Subject: Re: [9fans] commercial deployment
Date: Sat, 21 Jul 2001 10:53:44 +0100
Message-ID: <200107210953.KAA06572@localhost.localdomain>

> I've used Plan 9 as a development platform since the 2nd Edition came
> out in 1995.  It was used to develope what is now the Cisco PIX Firewall
> and the Cisco LocalDirector load balancer.

do you mean that Plan 9 was the host system, while developing the PIX
as a target system, or that Plan 9 was the target system _in_ the PIX?

steve

Re: [9fans] commercial deployment

[Steve Kilbane]

> I've used Plan 9 as a development platform since the 2nd Edition came
> out in 1995.  It was used to develope what is now the Cisco PIX Firewall
> and the Cisco LocalDirector load balancer.

do you mean that Plan 9 was the host system, while developing the PIX
as a target system, or that Plan 9 was the target system _in_ the PIX?

steve

Re: [9fans] commercial deployment

[bwc]

I've used Plan 9 as a development platform since the 2nd Edition came
out in 1995.  It was used to develope what is now the Cisco PIX Firewall
and the Cisco LocalDirector load balancer.  I think they changed all that
to Sun's and GCC after I left Cisco.

I am currently using it in my company.  Besides being the rightful heir
to the timesharing crown having a direct linage to CTSS, it's the
most pleasant environment for an old Unix hacker to work in.

  Brantley

[9fans] commercial deployment

[Stephen King]

I like plan9 philosophy very much (being something of a minimalist
myself) but I must ask (hoping it's not been asked a thousand times
before):

	Is there or will there ever be commercial/industrial deployment of
plan9?

I imagine distributed very-large-set processing as a viable commercial
application but will that ever really happen?

You opinion is welcomed, flames to /dev/null ...

--
"I think my life is fuller because I realize that I
don't know what I'm doing" - Richard Feynman, 1979.