Slashdot Mirror
Sun community licensing High Performance Cluster Software
Anonymous writes "Sun just announced that they are open sourcing their
high-performance clustering software." The announcement is on Yahoo. Sun will be releasing it under their Sun Community Source Licensing, which is different then XFS, which was truly Open Sourced.
In the 1970s I was doing system programming on a CDC mainframe. We and other customers customized and improved the OS, used CDC knowledge bases and electronic bulletin boards to share the info with CDC and other customers. Manufacturers also sponsored user groups which also often had magnetic tape archives of various software tools.
Of course by then the more well-known UNIX source code was permeating universities.
Now more manufacturers are rediscovering the benefits of giving away razor handles and selling the razor blades for it, even if it does tend to produce more discussion about the benefits of various brands of compatible blades...
The article originally read something to the effect of "HPC2.0 is being released under SCSL, just like XFS."
Christopher A. Bohn
cb
Oooh! What does this button do!?
Umm... that's right. In fact, that's exactly what it says. Read it again:
...releasing it under [SCSL], which is different then XFS, which was truly Open Sourced
In other words:
Sun's cluster stuff was SCSL'd. XFS was not. XFS is truly Open Source. The implication is that SCSL isn't.
It pays to reread and rethink. You'll spare yourself a misinterpretation now and then.
I like to play children's songs in minor keys.
"We're all sons of bitches now." --J. Robert Oppenheimer
I don't think he was trolling. Don't be so paranoid; the UCB/BSD style of licensing does have it's advantages. One thing I've always wondered that rabid GPL advocates continue to spout is how does a company "stealing" (another issue, you can't steal what is freely given away with no strings attached) something licensed with a BSD-style license diminishing the original code? It doesn't.
I think it would ne much nicer if they only released it as GPL.
BSD sucks then it comes to preserving freedom. And hey, *BSD trolls suck then it comes to discussions of weaknesses in your loved OS. If all you want is to help companies rip us off whty don't you support SCSL instead
Good lord, I hope not. Why would SGI think of limiting the inpact of XFS by placing it under GPL? I hope they consider some other less restrictive license that won't infect any works it touches.
Yes, I've installed Sun's HPC/LSF software on a machine here for a user who analyzes systems and compilers for performance, specifically Fortran 90 code.
We currently have it installed on a Sun Enterprise 3000, 6 processors (167 Mhz), 1.5 Gb memory, and in preliminary tests is outperforming our 128-processor SGI Origin 2000 (don't know cpu speed off hand).
We're currently only using it in a 1-node configuration, but the neat thing about the clustering software is that if you have to take a node down, the processes running on that node can be moved to another node. It's really quite a significant thing when you think about it - a process being stopped, it's entire memory space being physically shipped to another machine, and restarted from where it left off.
Bob Campbell
I'm not going back. I like things the way they are now.
Bruce
Bruce Perens.
Of course it doesn't diminish the original code, but the GPL forces any improvements to be handed back to everyone - not only to those who choose to pay for it - which BSD-style licenses don't. Furthermore you get to see the source code, which you can learn from. So in time, you might find that a commercial program derived from an originally free program to be better than the work it was derived from. If turns out to be the case free software users are in fact losing out.
Does anyone actually know anything about this product--worked with, read up on it, etc?
Wil
--
Internet Meta-Resources:
Wil
wiki
What is this viper that you speak of? I have never heard of it before (which doesn't mean much). Can you give some details and/or urls? gracias
Why? SGI could only lose money by putting XFS under a less restrictive license. For example, if XFS takes on a BSDish license, BSDI or SCO could sell a restricted commercial port to their platform, and make money off of restricting our uses of it. On the other hand, just because SGI releases XFS under GPL, it does not mean that they can't also release it under a different license for their own purposes.
What's the second easiest way for a big company to inspire hate within the linux community? Don't release any source code!
What's the easiest way to get the linux community to hate you? Release your source code!
No matter who you are or what you do, if you need to make money off your code, the Open Source People will blindy hate you, so kudos to Sun and others for doing it anyway. Hey, aren't we due for another round of Apple bashing about now?
HA HA
And when I get XFS on my Linux Ill start rooting publicly for them at every chance I get!
--
AC
You can't start it. This whole thread is nothing
but a licensing flame war. Not total war, perhaps. More of a cold war. Here's my constructive contribution: GPL stinks on ice.
Freedom does not mean doing what Richard Stallman
says. It means being responsible to your own
conscience.
give this man a 2!
-- your knees hurt, don't they?
It would be much nicer if they used an X or BSD style license.
No, it's not. It's about being free to extend software better to match users needs. It's about being free to take working, tested pieces of software and combine them in new ways. It's about being free to spend time on interesting problems rather than on yet another linked list implementation. It's about finally being able to make a round wheel and to use it freely.
In the short term, the one-and-a-half new pieces in their HPC release will benefit a handful of users. In the long term, there is no benefit to users in Sun's release. It's a dead end.
Jason, ejr@cs.berkeley.edu
>basically, the SGI guy wants to GPL it, if the lawyers will let him. (not bad, eh!) However
>they don't discuss other OSs - (I hope XFS won't be just for Linux).
If it's GPL'ed, then it would only be available to GPL'ed OSes such as Linux or Hurd. LGPL would be a better route if we want to be able to use it for proprietary- or BSD- licensed OSes.
I know that some of the confusion is the original story submitters, and some comes from Hemos' comments, but it's at times like this that I think the term Open Source is just as confusing as Free Software. People look at something where the source is available, and ask themselves, isn't this open source? In both cases, the words don't mean quite what they look like. Something that has available source code isn't necessarily Open Source; something that is no cost isn't necessarily Free Software. I don't see how the new phrase we have all adopted has cleared much up at all. What we're really after is source code we can modify, redistribute, not just read. Witness this Sun Community license as an example of where that confusion can come back to haunt us.
--
Ian Peters
So what license _did_ SGI use on XFS? Is there a link somewhere?
Actually, SGI people on the linux-kernel mailing list have said that all mods to the kernel proper will be under the GPL (obviously), but that the licensing terms for any modules has yet to be decided (their legal department is apparently hashing it out). I hope they release it ALL under the GPL, because the last thing the kernel needs is a smattering of almost-open-source-but-not-quite-so-its-not-includ ed-with-the-kernel modules. That would also encourage other companies to do the same.
No, Anonymous Coward, GPL does not prevent anyone
from integrating XFS with their OS. It just prevents them from distributing the resulting OS in binary format. It is trivial to work around
and practical restrictions this imposes on commercial closed-source systems (see numerous
existence proofs) and it just does not apply
to truly free systems such as BSDs.
A sly dig at MS? :)
Bruce, I've tried for several days to email you (bruce@perens.com) and it keeps bouncing on me. If you have working email please drop me a line. Otherwise checkout the archives at debian-legal (thread "A Data License").
-matt
For some reason when I read articles like this, I am confronted with an image of a light puffy snowball rolling down a gentle hill. It just seems to get larger and larger and faster and faster the further it moves along the curve. Where is it going? Down to the lake, the beach? Who knows, but it destroys everything that stands in its way.
+&x
That was to be Ross's 64 bit Ultra killer. It was apparently about 40% more efficient for the clock speed and was looking at 300MHz right out of the gate. Unfortunately, Ross ran out of time and money. Too bad. I have liked their modules for a while and have several quads in my back office running my database and accounting (and possibly SAP in a few months) and with 140MHz and 1MB of cache, they really fly, even with only 512MB RAM.
Sun does OK with SPARC, but Ross did better. I remember when I was thrilled to get 125MHz units and Sun was still shipping 33MHz units. But Ross never was willing or able to bring the prices down to a level that could compete and Sun certainly gave them no price breaks for the OS licence, so that really kept them from being the Dell of SPARCs, which they could have been. And then the Ultras really took a lot of their market, despite the fact that Sun's 64 bit support has been pretty flakey until recently, to say the least, and I think that without their desire to capitalize on the old Cray stuff that they bought off of SGI it still would be.
Had Fujitsu kept Ross alive, I expect that we would be seeing 600MHz modules with 16MB L2 from Ross, blowing away Suns. And, in the finest Ross tradition, running so hot that you could warm biscuits on the heat sink (yes, it is a good thing that the executives don't see everything that goes on in the machine room in most companies)(I don't want to know what goes on now that I have my own!). I am still interested in who got the IP from the deal. Anybody know?
You misunderstand. the linux community will
find you distasteful if you try to make money,
like redhat. the quickest way to get readers
of slashdot to hate you is to release your source
code, pretending to be friendly to the community,
but actually the code is still propietary.
the way to make the linux community love you is to
make high quality truly free software. whether
or not you happen to make money at the same time isn't much of an issue. the only reason redhat is hated is because they attempt to use marketing power to negatively influence standards.
Hamsters are at least as feathery as penguins. HamLix
you can't expect every company on the planet to completely give up control of their technology. how much do you think sun spent on making their clustering solution, which has availability features as well as distributed processing features. (read some later posts)
The benefits of sun making this code public outweigh whatever bruce perens thinks about whatever. If the other UNIX companies follow would follow suit it would be a real victory for *our side*. (our side being customers, administrators, and users. not techno evangelists, ceo's, stock holders, or "religious fanatics".)
The truth is that the commercial unix people who have spent huge sums of money on making the Solaris's and AIX's of the world what they are today. They should move to a community development model so that they could compete in the commodity business of server software that is already here. It irks me that as Linux became more popular you started seeing the long time open source people calling it GNU/linux and the like. The success of linux is the validation of some of their ideas, not of them or their agendas.
It's all about usable software in the hands of users. The rest of this is pointless.
Some mails from the SGI people to linux-kernel say
that it will be licensed under GPL.
I always thought that Beowulf was just a setup involving rsh (remote shells) and not true SMP clustering which is what this thing from Sun purports to be isn't it? .. Like just run something on the cluster and the system figures out where it runs on. .. hmm.
/.'er will correct me :)
... lots of cheap computing power.
I read on NASA's site somewhere about the 'Hive' that runs Beowulf and they said there that it's all based around rsh and also runs out of process table space too
If I'm mistaken I'm sure a great many
Clustering IS a lovely idea for Linux
Delphis
I think that was supposed to be "the same as the Java 2 platform".
Christopher A. Bohn
cb
Oooh! What does this button do!?
AMEN MY BROTHER
Regarding Apple-bashing, Apple very nicely incorporated all of the points that the Debian folks and I raised into version 1.1 of their APSL license. Whoever was Apple-bashing, it wasn't us. We had constructive criticism, and Apple responded to it. The press saw it as an attack, but they weren't used to free software issues.
Thanks
Bruce
Bruce Perens.
Posted by Hank Shiffman:
Just one correction (as the guy from SGI who was quoted by Linux World): XFS is a 64 bit file system, meaning that it uses 64 bit offsets everywhere to allow lots of space for files. It does not require a 64 bit OS; XFS is supported on our O2 systems which run a 32 bit version of IRIX. Oh, and as I read the terms of Open Source licenses it's clear that we don't get to decide that XFS is only for Linux. Not that we would want to discriminate against FreeBSDers or other deserving souls. Heck, if Sun wants to GPL Solaris we might even enjoy having them use it!
Please fix this article. "Sun Community Source" isn't Open Source - I think you went over their press release too fast. It's also not the same license that SGI used on XFS. SGI's license appears to be Open Source and they are being a lot nicer to the community than Sun.
Thanks
Bruce
Bruce Perens.
Bear in mind that the Slashdot people quite often change the text of headline articles in response to corrections. It may not have read that way when Bruce posted his comment.
--
Xenu loves you!
Just so happens I'm getting involved with some colleagues and the company I work for in Beowulf style applications. We went down to Linux Expo to check it out, and it was really good stuff. Our plans to front-end with Sun and back-end with Linux has just gotten a whole lot more solid - not to mention cooler. Way to go Sun!
Ok, for clarity:
SGI is releasing XFS under some yet-to-be-disclosed open source license
Sun is releasing some HPC stuff under their "Community License." Sun has not claimed anywhere that this constitutes "Open Source." "Open Source" is not mentioned anywhere in their press release.
See this essay
Bruce Perens, now there's a man who has every right in the world to decide what is Free and what is not (sarcasm).
Why does everyone become squeemish and unable to think for themselves when it comes to the issue of Freedom? Why? Because the GNU/Linux community has lost it's moral compass. We need to return to the old ethics of freedom, the whole freedom, and nothing but freedom. OSI/Open Source(tm) is corrupt at its foundation - it was formed to act as a "cleaned up" hacker group. I, for one, find nothing wrong with the hacker community in it's raw, untidied glory. Relax and be comfortable with who you are - do not follow the Pied Pipers like Bruce Perens and Eric Raymond. To yourself be true, only to yourself.
The term "freed software" should be used instead of "free software". It's unambiguous and no one could confuse its meaning. What do you think?
Freed software = free software > open source > proprietary software.
Right. Linux need high availability; it is fine with high performance (yes, I know FreeBSD is better, now put the flamethrowers away). The link for what Sun could really help us out with is:
...
http://www.sun.com/clusters
That would be cool.
To be honest with you, I tend to find that Sun's behavior is a little puzzling. Is this to try to distract from Beowulf? I realize that having more code to look at helps people wrap their minds around a problem, but I don't really see where Sun gains here. Perhaps if Sun would cut the cost of their better CPUs (I want a 400MHz UltraSPARC with 8MB of cache, but not for $4000) a little and do the same with their rebadged commodity hardware they would lose less to Linux on X86. SPARC is better. SPARC is cool. SPARC costs too damned much. What a pity that Viper never saw the light of day
Ideally, we could have a Beowulf cluster with HA funtioning within the cluster and apps running over (I am not sure of how this works with Mosix) it all. Give me Workload Manager and CICS (and ISPF!!!) for Linux and I would be a very happy camper. As far as I am concerned, the glass is now half full (Mosix, Beowulf) and rising.
Yes, you're right; these are both compatible with the GPL and would permit use under a broader ranger of circumstances than would a straight GPL license. But let's not start a licensing flame war. :-)
Bruce
Bruce Perens.
I'm pleased to see Sun providing source for more products. Albeit they are not using the GNU license, the fact is that their products are becoming more open to peer review. The goal isn't for the GNU license to pervade the software cycle completely. It has influenced it in a positive way, and the progress is exciting.
-- Solaris Central - http://w
This is high performance clustering like Beowulf. Linux needs high reliability clustering, which this is not.
Some of the more interesting pieces, like LSF, are only licensed by Sun, thus will not be included in this `deal.' (For a free improvement over LSF, check out GNU Queue. If it doesn't do something you want, you can support the community and extend it.) If you read the announcement carefully, you'll see that the only new codes to which it applies are the parallel file system (the Sun CTO thinks distributed file systems are dead, anyways), the Prism debugger, and the parallel run-time environment.
Of those, the only with no available substitute is the debugger. The ROMIO library is a good place to start for the MPI file I/O stuff (a good database would be a better place, imho). I already mentioned queue management software. The Ptools Consortium and the Globus Project have links to other HPC cluster tools.
Many of the pieces for debugging are available (combine ddd and gnuplot), but some notable ones are missing. The ability to control multiple GDBs easily from one processes and the visualization of parallel execution are needed, and quite difficult to implement. There seems to be interest in making GDB easier to use from other processes, which is a good start towards solving the larger problem of general, distributed debugging. And both the mpich and LAM MPI implementation have some profiling information, but few tools to dig through it.
To be fair, Sun has contributed (and supported contributions) to the original packages. Why they are releasing the rest under their Exploit the Community license is beyond me.
Jason, ejr@cs.berkeley.edu
btw, Sun's SCSL is aimed more at commercial developers (including Sun's OEMs) and researchers, not so much general members of the public. However, they are releasing quite a bit of stuff under the SCSL - Java, Jini, HotSpot (later this year), their SPARC processors and several other software products. They seems to be SCSL'ing their products in general. They haven't said much about SCSL'ing Solaris recently - the last time it was brought up they said it would be quite hard to do, because of all the liscences.
I suppose there will be inevitable comparisons between Beowulf and Sun's HPC software, and SMP kit. The main hardware difference is bandwidth and latency - Beowulf seems more about combining lots of single CPU (or low CPU count, eg 1-4) boxes in a network, possibly having several hundred of such boxes. Sun's approach to high end computing is to have big SMP boxes (a single Starfire E10000 can take 64 UltraSparcs) with the option of clustering a few of them - currently limited to 4, ie 256 processors. A Starfire has a 6Gbyte/s I/O bus and 15Gbyte/s main memory bus, which is rather better than Ethernet. Sun's approach is more expensive, but it also solves a wider class of problems well. For some things (eg cracking codes, rendering) you don't need much interprocess communication or bandwidth, so it scales well with Beowulf, but for other things (some kinds of database operations, eg OLAP, and data intensive scientific calculations) you really need very high bandwidth and very low latency (close to main memory speeds) which is where Beowulf doesn't do so well. Still, some things don't scale so well, even on a Starfire... Btw, the Starfire is over 2 years old.
Cue Sun's next gen super-computer, codename Serengheti, which has a completely different architecture. It's memory architecture is called Cache Only Memory Architecture (COMA), which seems to have been in development for a long long time at Sun. A single box will take 128 processors, and you'll be able to cluster 8 of them, for a total of 1024 processors. It'll be powered by Sun's UltraSparc-III, which recently reached first silicion, and has b ooted on Solaris. Incidentaly, the UltraSparc-III has hardware support for 1024 processors, and is supposed to be out in volume production by the end of the year. However, Serengheti won't be out until about the 2nd half of 2000.
It is worth pointing out once more that this DOES NOT mean that Linux will get any benefit from this development.
The Sun Community Source License allows you to see the source, but does not allow general free use or redistribution. This is not some picky "Oh it's not GPL, Ah it is not fully Open Source (TM)" point. It just isn't a free use or free redistribution license. It allows for research use, paid-for commercial use, and redistribution only among existing licensees.
The release has technical interest, and I'm happy that Sun have done it for reasons other than to give a false impression of contributing free software, but it's of little use to most of us. It's only of use to Sun platform developers and academics. It may be very good news to Sun's customers using the affected products.