Plan 9 Running on Blue Gene
gholmer writes "Eric Van Hensbergen reports that Plan 9 has been successfully booted on IBM's Blue Gene supercomputer. A live demo will be attempted during a poster session at this year's Usenix. There is also the obligatory Space Glenda picture."
Can it run Vista?
What if the Hokey Pokey really is what it's all about?
"Plan 9?
Ah, yes. Plan 9 deals with the resurrection of the dead. Long distance electrodes shot into the pineal and pituitary gland of the recently dead."
That's pretty cool to think about since I didn't know if Plan 9 is still used or not beyond research environments. Or even developed for. Also I was beginning to think Slashdot was dying since I hardly come here anymore, but with news like this I feel mistaken.. Pretty cool!
Am I the only one that was wondering why Plan 9 Pulishing (http://www.plan9.org/) would need to be run on Blue Gene?
Is this a going concern? Do lots of people use this?
What happened to Plans 1-8? And could you make a module that corrupts the output, and call it Plan B? I think it may be a little too early to grasp exactly what the story is here. Where's my caffeine?
Forget Vista, can it cook toast? Without burning it?
This sig left intentionally blank.
I'm not buying a Blue Gene until they port AmigaOS to it, like God intended.
You are not alone. This is not normal. None of this is normal.
Stupid! Stupid!
When you've got solamanite, you've got nothing!
(Yeah, its one of my fave movies)
"Enjoy what you're doing! If it becomes drudgery, you're doing it wrong!" - Jim Butterfield
Sorry but it's not really clear what it all implies. Could someone explain?
You just got troll'd!
The graphical programs displayed are: the mail announcer faces(1), the system statistics watcher stats(8), the text editor acme(1), the sky catalog scat(1), the image viewer
I'm not sure I'm ready to check out any "graphical" items called scat.
That is, if you look at the source. Man, for such a simple page they sure use an enormous amount of tage. If this is what the future looks like, I want the past back !
Religion is what happens when nature strikes and groupthink goes wrong.
Considering some of the low power hardware that NetBSD has been ported to, I am sure that a top of the line IBM super computer should have no trouble handling it :)
Don't judge a book by its cover. The current generation of flashy-looking OS's are excellent for computers with a small number of CPU cores and uniform memory, but they are really poor for machines with many cores and core-local memory. Plan 9 is designed to work as a distributed OS, which is perfect for Blue Gene, and it will probably become more and more relevant to home computing as we move towards PCs with thousands of CPU cores, because we'll need a decent distributed OS to make use of them. The mid-80s "FVWM" look is just because it is a research OS and the researchers have better things to do than port KDE.
http://plan9.bell-labs.com/plan9/download.html
It's been released, and is on the Fourth Edition.
With the added power of Bluegene, the Bell Labs team will now be able to add more then on colour to their GUI.
"Well , cutting edge for 1990. If thats the best it can do on a supercomputer it doesn't bode well for your average PC!"
Super computers don't run GUIs. That is for visualization workstations.
"Has it broken any new ground with any new operating paradigms? (Thats a genuine question , I don't know)."
Yes I suggest you go learn a lot more about it before posting in blatant ignorance.
Plan 9 is a distributed operating system. It uses clusters of servers to act as application servers, storage servers, and IO servers. It is ideal for clustered systems with hundreds or thousands of cores! Guess what Blue Genie is?
Supercomputers usually lack a traditional gui. They depend on workstations to handle any visual interface. They are all about speed and nothing else. Your comment about a less than pretty GUI on a supercomputer is about as useful as complaining about the crappy stereo in a formula one car.
Is Plan 9 important? Well since it looks as if cores are going to start multiplying at a Moore's law like rate then the answer is most likely yes.
See my blog http://ilovecookes.blogspot.com/ for light hearted technical information.
"Has it broken any new ground with any new operating paradigms? (Thats a genuine question , I don't know). I do wonder why thety bother and don't just try and integrate any new ways of thinking they've come up with into pre-existing systems such as Linux or BSD."
Well, yes. Read the overview
-mls
The WHOLE POINT of Blue Gene is to do intellectual exercises. It's a RESEARCH computer.
Currently hooked on AMP
Blue Gene is a very specialized supercomputer designed with a customized 'OS' (if you can even call it that!) which minimizes any sort of interrupts and other nonsense such as typical OS stuff because when you're scaling out to 65,536 nodes on an MPI-based code which requires lock-step synchronization, you can't afford for some unimportant process on a single node to cause small delays. Plan 9 IS a research oddity on the system in this regard, and not the sort of thing you'll see anyone putting on a BG/L for what it was intended to do.
(This doesn't mean you won't see it eventually if someone has way, way too much money to burn - after all, the PS3 is designed for games, but some people are experimenting with them for computation - but let's not get carried away. The point is, BG/L is not the sort of system that Plan 9 would be targeted at.)
"your tourettes-like thoughts to yourself "
You were the one who swore sonny , though I suspect you don't even know what Tourettes is but you just heard it somewhere and thought it sounded cool.
"Perhaps while we're doing that, you could follow some of the links on the article and educate yourself about Plan9?"
Been there done that. I still don't see its point.
...turn it into a Sinclair ZX81 to prove that it can be done?
What's next? ReactOS on a Cray?
Politics is the art of looking for trouble, finding it everywhere, diagnosing it incorrectly and applying the wrong fix.
"It's a RESEARCH computer"
Its for researching problems using a computer. Its not generally for research computing issues themselves.
I do wonder why thety bother and don't just try and integrate any new ways of thinking they've come up with into pre-existing systems such as Linux or BSD.
Why would they put a 16 year old consumer-oriented, x86-based, single processor-optimized operating system on a distributed supercomputer? I dunno, maybe they're just a little dim.
The whole purpose of the project is to research new ideas which make their way into production operating systems. "Slowly, ideas from Plan 9 are being adopted by other systems. Plan 9 was the first operating system with complete support for the UTF-8 Unicode character set encoding. The dump file system has been mimicked in Athena's OldFiles directories or Network Appliance's .snapshot directories. The flexible rfork(2) system call, the basis of lightweight threads, was adopted as is by the various BSD derivatives and reincarnated on Linux as clone(2). The simple file protocol 9P has been implemented on early versions of FreeBSD and current versions of Linux."
Plan 9 is a radically distributed OS. It was written from conception as a distributed kernel, and all aspects of the OS are distributed in ways that Linux/Unix/Windows are not. It may be older, but it embraces many distributed paradigms that few OS's in production can handle. Because it is so distributed, the many common utils are simply not compatible with the kernel without a ground-up rewrite. Emacs Emacs, X, KDE, Gnome are not ported and probably won't be. Here's a naive review: http://www.osnews.com/story.php/15235/Investigatin g-the-Plan-9-Operating-System
only in the proprietary software world do you need fat software for fat hardware (or vice versa).
Thin software running on burl hardware frees up your resources, gives you room to really flex your muscles. That's what make those 386-linux guys l33t. They get a 2.2 kernel and a copy of busy box, throw it on a 15 year old machine, and when they hit the terminal they still have leg room.
Under the influence of Post-Cyberpunk Gonzo Journalism
I'm sure Plan 9 is an interesting intellectual exercise for the people involved, but other than that , what exactly is its point?
The point is to make distributed computation a whole lot simpler than it is right now.
I do wonder why thety bother and don't just try and integrate any new ways of thinking they've come up with into pre-existing systems such as Linux or BSD.
They do try, and they have succeeded to some degree.
But that's fraught with its own problems; for example, few if any Linux programs will know anything about 9P or naming or any of that.
last I heard, neither were viruses.
-- 'The' Lord and Master Bitman On High, Master Of All
Its for researching problems using a computer. Its not generally for research computing issues themselves.
What is the difference? Is computer science not "science?"
Plan 9 looks to me like the perennial contender for something which is never to be released; much like the HURD
Plan 9 has been released, and it's working.
Is it going to catch on? Who knows. It took 20-30 years for UNIX to catch on after it had matured reasonably well, so that would put Plan 9 taking over the world at somewhere between 2010 and 2020.
And you can run it using Xen, so everyone can try it out and see what's it about.
8 of 13 people found this answer helpful. Did you?
But BG/L is exactly the type of hardware that Plan 9 is designed for.
Plan 9 could allow Blue Gene to be used for different problems than it is currently being used for. Yes it is currently are research project but it is far from a waste of time.
I disagree that BG/L isn't the type of system that Plan 9 is targeted at. The current problem set that BG/L is being used for isn't one that Plan 9 is a good tool for. The hardware probably isn't ideal but it is close enough for useful research.
See my blog http://ilovecookes.blogspot.com/ for light hearted technical information.
Do people really need to say "non-viral"?
Yes. It is a genuine concern for many people.
This comment is fully compliant with RFC 527.
When I log into my Xenix system with my 110 baud teletype, both vi and Emacs are just too damn slow. They print useless messages like, 'C-h for help' and '"foo" File is read only'. So I use the editor that doesn't waste my VALUABLE time. Ed, man!
/bin/ed /usr/ucb/vi /usr/bin/emacs
!man ed
ED(1) UNIX Programmer's Manual ED(1)
NAME
ed - text editor
SYNOPSIS
ed { - } { -x } { name }
DESCRIPTION
Ed is the standard text editor.
Computer Scientists love ed, not just because it comes first alphabetically, but because it's the standard. Everyone else loves ed because it's ED!
"Ed is the standard text editor."
And ed doesn't waste space on my Timex Sinclair. Just look:
-rwxr-xr-x 1 root 24 Oct 29 1929
-rwxr-xr-t 4 root 1310720 Jan 1 1970
-rwxr-xr-x 1 root 5.89824e37 Oct 22 1990
Of course, on the system I administrate, vi is symlinked to ed. Emacs has been replaced by a shell script which:
1. Generates a syslog message at level LOG_EMERG;
2. reduces the user's disk quota by 100K; and
3. RUNS ED!!!!!!
"Ed is the standard text editor."
Let's look at a typical novice's session with the mighty ed:
golem$ ed
?
help
?
?
?
quit
?
exit
?
bye
?
hello?
?
eat flaming death
?
^C
?
^C
?
^D
?
---
Note the consistent user interface and error reportage. Ed is generous enough to flag errors, yet prudent enough not to overwhelm the novice with verbosity.
"Ed is the standard text editor."
Ed, the greatest WYGIWYG editor of all.
ED IS THE TRUE PATH TO NIRVANA! ED HAS BEEN THE CHOICE OF EDUCATED AND IGNORANT ALIKE FOR CENTURIES! ED WILL NOT CORRUPT YOUR PRECIOUS BODILY FLUIDS!! ED IS THE STANDARD TEXT EDITOR! ED MAKES THE SUN SHINE AND THE BIRDS SING AND THE GRASS GREEN!!
When I use an editor, I don't want eight extra KILOBYTES of worthless help screens and cursor positioning code! I just want an EDitor!! Not a "viitor". Not a "emacsitor". Those aren't even WORDS!!!! ED! ED! ED IS THE STANDARD!!!
TEXT EDITOR.
When IBM, in its ever-present omnipotence, needed to base their "edlin" on a UNIX standard, did they mimic vi? No. Emacs? Surely you jest. They chose the most karmic editor of all. The standard.
Ed is for those who can *remember* what they are working on. If you are an idiot, you should use Emacs. If you are an Emacs, you should not be vi. If you use ED, you are on THE PATH TO REDEMPTION. THE SO-CALLED "VISUAL" EDITORS HAVE BEEN PLACED HERE BY ED TO TEMPT THE FAITHLESS. DO NOT GIVE IN!!! THE MIGHTY ED HAS SPOKEN!!!
?
Okay, with a name like that I definitely clicked on the link... and I feel *so* cheated.
Want to improve your Karma? Instead of "Post Anonymously", try the "Post Humously" option.
To some of us yes, its important and does factor into decisions as it can cause long term ramifications.
If licensing restrictions didn't matter to people, we wouldn't even have the concept of BSD license to discuss ( or GPL ), would we?
---- Booth was a patriot ----
Uh, never mind...
MSBPodcast.com The opinions expressed here are my own. If you don't like 'em... Think up your own stuff.
So I suppose it will replace every UNIX system by 2038?
Is where IBM raises the dead in an attempt to takeover Earth without ever having to leave the mother ship.
"He who can destroy a thing, controls a thing." --Paul Atreides, Dune
I was talking about Linux or BSD. Not Windows.
The Plan 9 license has been a struggle with Lucent lawyers. Plan 9's 3rd edition was the first that was offered as a free download (which is where I found it via /.) At this time it contained a clause saying you had to submit any kernel changes you made back to Lucent. The Plan9 team fought to get rid of that and did a pretty good job of releasing it under what was thought to be an open source license. The OSI and RMS thought different, iirc it was because of the US munitions export license problem. The license was redone and approved by the OSI as the license we see today.
There was a time where there was a tounge in cheek tickbox that said something like : I promise not to use plan9 to make nuclear weapons outside of the US.
Licensing arguments were pretty common on the mailing list as though the devs were not trying to get it sorted or were in control of writing the damn thing. It got distracting enough that licensing discussions got their own mailing list, which then we could all not bother reading.
There are places where the networks are not touching,and there are places where they are-Boeing's Lori Gunter
Plan 9 was designed at Bell Labs as the successor to Unix. Its primary characteristic is that EVERYTHING is managed as a file, down to devices. So if you have a CD in your drive, and you only wanted the data track of the CD mounted, you'd delete the subdirectory containing the audio track and it'd be unmounted. It never really caught on outside of research environments.
Hopefully plan9 will catch on by the year 2020. that will give us 18 years to phase out unix before Unix EOL's itself in the year 2038.
i thought once I was found, but it was only a dream.
What you say is not true.
/proc filesystem and the like. In practice, I think having resources controllable as files is useful only to the entent that nearly every programing language has fwrite() and fread() capability in some way, even if just by shell redirection. There is actually no point in being able to open up a file editor, save to /dev/wifi and have those bits go out over the antennea.
I worked for a time with some people who were deep 9 fans. They wanted to build a network of wireless access points running Plan9, and a big "computing cloud" or "resource cloud". They thought you could make a system where you logged in at any place, and all your windows popped up there with your desktop as you left it halfway accross town; your processes would supposedly migrate from CPU to CPU according to how best to allocate resources; all the files would be spread accross everyone's disks so no one disaster could loose anything, but they would migrate close to you for speed as you used them; etc etc. Basically every cool sounding impractical idea you ever heard of from Freenet to sci-fi cypercrap they said you could do with Plan9.
I got on the 9fans list and read a lot of the web site and then got to work. I found out that they had this concept of a cpuserver versus a disk server versus authentication server, so I had to buy two machines; running everything on one machine was possible but somehow more complecated (I think that is what most people do now, and what the Xen image does, though). So I bought 2 machines and built a third from cast off parts and started installing. I took careful notes of everything, when stuff didn't work I would wipe the disk and start over taking notes again.
I figured out the following things:
1) There is no migration of processes from one CPU server to another. When confronted, the hypesters said "well I never said we wouldn't have to write code"
2) The disk server isn't anything more than any networked filesystem with authentication. They kept making a big deal about the great ideas in their filesystem, and talking a lot about some crap that they had written back when they had a big jukebox of CDRs that made them act like a big mutable filesystem and tracked changes. In the end, it is all just files and directories and passwords, there is nothing new; no files migrate to the machine with the faster disk if they are used often or anything like that. In fact it has bugs and "documentation bugs", but they have so few people using it and helping each other on 9fans that often bugs never get established as being real bugs with the people who should fix them.
3) The user interface is horrible. I think it was written by some guys at Bell Labs who never had a graphical computer in their lives, saw windows 3.1 taking over the world, so read some theoretical papers from SIGRAPH or something and cobbled something together. I have a strong suspicion that Rob Pike and those other fellows can't touch type, and operate a computer by hunting and pecking with the left hand while the right hand operates the mouse. That ACME editor and the interface start to make sense if you immagine a Rip Van Winkle stuck in Bell Labs since 1960 peering out at the world briefly about 1984, and then scuttling back inside and madly hacking some crap for a decade, and then producing it in 1994. If you read the mailing list, it becomes plain that most of these guys use Windows and connect to their Plan9 fetish boxes using a remote desktop tool. This is the only way they can browse the web. All except for a few don't even read their email on Plan9, which supposedly has a great tool for that, they use outlook -- just check the headers on the mailing list.
4) that scripting language rio is just another sh
5) they make a big deal about an "everything is a file" paradigm, parts of which were copied into unix as the
and lot more of the same.
As I worked through these problems and posted on 9fans, I slowly started
If I had been talking about Windows, I would have said a 22-year old operating system.
What about "viral marketing"?
There nothing negative about this phrase is there?
There were about 30 people attending the International Plan9 Symposium in Madrid last year (of which I was one). So how does that compare with *BSD?
Because if someone, post-1995, released a new, general-purpose OS that wasn't Unix-compatible, everyone would laugh themselves silly.
If you are a Unix system, you are part of a family -- there are gigabytes of useful source code to port, and there are plenty of nice ideas to work from or expand upon.
If you're not a Unix, you are alone. At least until someone takes pity on you and writes a Unix compatibility layer ...
Yes, and this is the only interesting story out of the 10% that haven't been on digg first.
I doubt Plan9 is digg material. (I take that back - 2 diggs..)
They'd like Glenda well enough, but who doesn't?
I think there will be very few 32-bit processors left outside of embedded systems in 2038. My system currently uses 64-bit time_t, and I assume most others do as well nowadays.
Centralization breaks the internet.
Well it does work on 486s.
And while it's been ported, every port is going to bring a certain level of unnecessary bloat and inappropriate design decisions. When you're spending millions and millions of dollars on a supercomputer, why skimp on the operating system?
It's one of the best distributed research OSs there is. We'll have to see, now, if it is as useful a research tool as hoped with so many processors.
I don't believe so. Interactions with hardware (and all I/O, for that matter) are inherently non-functional, since input is not guaranteed to be the same on every run. It doesn't make any difference whether you call it a file or something else. Monads seem to be required either way.
It's good to see that you guys "who hardly ever come here any more" happened to be here on the same day as this announcement.
like Ab Initio's Co>Operating System. It uses distributed file systems as well for distributed Extraction Transaction & Loading of data warehouse type applications. But it's as expensive as hell, like $5-10k per processor licensing fee. Be interesting if something like that was built on top of this.
http://blog.slaingod.com
L/unix and family and Windows et all have similar perspective; one computer/cpu per user possibly connecting to a network. SMP is an after thought and hack, the programming has to be rethought.
Plan 9 alters the perspective of the computer and how it interacts over a network. I understand it to work that every Plan9 site acts as though all parts of the network are available to each user. The basic idea is that the more resources on the network the better your computing experience. Also Plan9 takes Unix's 'everything is a file' even more literally. Including the window manager. kinda freaky to use, never got used to it so have given up for now. Maybe version 5 or 6 may be more 36pack joe friendly.
Semi-automatic amateur armchair Australian philosopher; conjecture ready at any moment...
It is getting better at chess, it takes more concentration each time, but still I win.
"And while it's been ported, every port is going to bring a certain level of unnecessary bloat and inappropriate design decisions."
And Plan 9 doesn't suffer from this? It was designed specifically for Blue Gene was it? I don't think so.
No, but it was designed specifically for distributed computing.
Gnu's Not Unix Plan9 Or Open Source now featuring the herd of cats kernel.
Laugh damn it, it's a joke.
Tired of all the isms, don't exploit people as an employer, or a government, mmmmK?