A UnixWare That Can Run Linux Apps 73
rafa writes: "Caldera seems to have some interesting plans for UnixWare, the operating system they acquired from SCO. Using SCO's Linux Kernel Personality software, they can run Linux software on UnixWare. This might also be used in AIX 5L, according to Ransom Love." The Caldera folks have been talking about this for a long time; what remains to be seen is whether enough customers are interested in the hybrid commercial / Free software system to pay the premium for it. The article quotes Ransom Love as downplaying the touted features of 2.4, saying, "It will probably take another three years to build a [truly enterprise-ready] Linux kernel."
Re:Enterprise-ready (Score:3)
The fact that so many Linux users can remain so ignorant about the system which Linux ripped off is astounding.
--
Ellison: How are you gentlemen !! All your database are belong to us
This has been in BSD for a long time (Score:1)
Dunno what the big fuzz is here.
- Hubert
It will probably take another three years to build (Score:2)
and what he really meant to say;
"It will be three years before linux has all the wizards that windows has"
Re:Like Debian/FreeBSD? (Score:2)
Why wouldn't they just recompile the GNU userspace to run natively?
No, this is an incredibly shrewd move to allow UnixWare to run all of those closed source binaries that are built for Linux like... umm... well I can't think any right now, but I'm sure there are loads.
Oh wait, I got it! Quake!
The Bible is not my book, and Christianity is not my religion.
Re:Another dead Unix (Score:2)
I don't know what it's like in the States, but in Australia the only commercial Unix that is alive and kicking is Solaris. HPUX and SCO are on their death beds. Digital Unix, AIX and Unixware are nearly unheard of.
Here in the states, I work in a shop that is about 50/50 split between HPUX and Linux, with a smattering of FreeBSD and Solaris (well, and file/print sharing is mostly NT, but I hardly ever have to get anywhere near it). It used to be that HPUX ran all the workhorse applications, but recently we've been moving some pretty large web/database applications to Intel/Linux. The HP9000s aren't going away anytime soon because, from a hardware perspective, they're more mature, reliable servers in a lot of ways. But as an OS, I think everyone has been a lot happier with Linux (even the BSD guy who still complains that he can't do a 'ps -ef' in Linux).
Overall, I agree with you that Solaris is the biggest commercial Unix by a large amount. Though there ARE plenty of people out there using that Sun hardware to run Sparc Linux...
As for Unixware, I'm not sure I see the point of Linux compatibility, but I do have a certain respect for Unixware as an OS. I admined a few Unixware boxes back when it was still a Novell property, and it was a good, by-the-book SYS5 implementation with a decent packaging system added on. This was back around '95, when most bosses would just sort of give you an amused look if you suggested running any sort of production server on Linux.
Re:Enterprise-ready (Score:1)
Personally, I think Linux 2.4.1 is as enterprise server ready as you can get, but I can see how UnixWare's file support and advanced I/O could make it even nicer--especially if you were running massive DBMSs.
For more on Caldera's UnixWare/Linux plans see:
http://www.zdnet.com/sp/stories/news/0,4538,268
Steven
Sun & SCO & IBM & IBM for the enterprise (Score:2)
> 64GB of RAM, a and multi-terabyte SAN.
I suspect this is not the definition used by SCO when describing the advantages of UnixWare.
In any case, IBM can deliver Linux based solutions with that kind of power, and better support.
Existing Unixware is bigger than you think (Score:2)
There's also a huge installation base of Unixware systems that Caldera would like to hold on to. These are mostly telephone embeded systems. When AT&T was a monopoly, computerized switches and voice mail systems were mostly Western Electric 3B systems running Bell Labs Unix. Lucent inherited that business, and even though they're not a monopoly, they're pretty strong. The systems now run commodity processors, but the OS is still Bell Labs Unix -- or rather its commercial successor, Unixware.
Do you have an AT&T or Lucent phone on your desk? Audix voice mail? Believe it or not, there's at least one Unixware box in your building.
Now, I don't know how tied Lucent is to Unixware. But they must be a possibility, however remote, that they would switch to Linux: no license fees, bigger developer community. Plus Lucent has competitors, some of whom do use Linux. Caldera needs to offer Linux compatibility just to hold onto all those license fees they get from Lucent.
__________________
Re:Caldera (Score:1)
Re:Enterprise-ready (Score:2)
Keep waiting! (Score:1)
So how many years will it take to build a truly stable Linux kernel? Six? Seven?
Re:But... what is the point? (Score:1)
If you're an ISV selling a niche product or products, you pretty much get to say "Customer, you will run SCO" or "Customer, you will run Linux". They just want you're application, they don't care what it runs on. They're paying $$$ for it to carry out a, presumably important, function, where there's little competition,
If it's an established, generic product, like a database server, you'll have a SCO port anyway. If not, you probably don't care about them.
If it's a new, mass-appeal product, like an RDBMS or a web server or something, OK, there might be some people running on SCO. I doubt it. Oracle doesn't seem to do SCO, for instance. But I don't think this is really 'news for nerds'; more like 'news for one proprietary Unix vendor who might be able to scrape two or three more years of life from it's dying product."
I don't have much experience of SCO, my company used to use it for telephony but it's gone now because the software provider was phasing it out; we now run NT. But everything I've read suggests it's used in niche markets, like telephony, where the vendor says "This runs on SCO, take it or leave it."
Re:But... what is the point? & UKP (Score:1)
Re:Like Debian/FreeBSD? (Score:1)
Exactly. Not user apps, though--server utilities. A company like Oracle would love to reduce the amount of time they have to invest in porting to different Unices... and more and more, these days, Linux is among the first ports that get done; IT managers ask about it a lot, whereas UnixWare hasn't built up much mindshare yet.
Well, now the Oracles of the world can port their stuff to Linux, and it'll Just Run on UnixWare--but it'll run better, from a gigantic-datacenter point of view, because UnixWare's multiprocessor and clustering support kicks Linux's ass on equivalent hardware. (Meaning no disrespect at all to Linux, which kicks UnixWare's ass in lots of other ways, but this sort of thing is exactly what UnixWare was developed and optimized for, and for all its flaws it's got an excellent kernel.)
Re:Linux needs more game developers (Score:2)
Re:I can only say... (Score:1)
Linux has DLL problems too, but we call them library versioning problems. Ever try upgrading your glibc version? Notice all the dependency problems when you try to do an apt-get dist-upgrade. There must be SOME reason that it held back a lot of my X font packages. Maybe its VERSIONING ISSUES.
Oh, my, 54000 bugs in an operating system with millions of lines of code.
Now, lets add up all the bugs in all the pieces of free software that microsoft would consider part of the operating system. That's kernel + x + a handful of servers + window manager + file manager. I bet it's darn near 54000 total if not over.
I've been running Win2000 since I installed it three weeks ago without rebooting for anything but driver installs. No blue screens thus far. And I abuse the hell outta my machine. AND I'm running beta drivers (video and dvd).
And it is a shame how much Microsoft is trying to emulate the Mac look-and-feel. Cause, I mean, they have a Desktop. And Windows. And menus and toolbars. And translucency. Just like a Mac.
Unlike linux, which has a Desktop. And Windows. And menus and toolbars. And is still working hard on ironing out the translucency. AND HAS AN AQUA THEME FOR GTK/KDE/GNOME/etc...
I also run Debian. I don't like Microsoft. But I hate people who troll with this kind of drivel. Linux is perfect and Windows sucks. Wait, that's a complete lie. Windows is GREAT at some things. Linux SUCKS at some things. In other words, they both have their strengths and weaknesses. Linux is not perfect. FAAARR from it. So is Windows though. Life with it. Open your eyes. Stop using slashdot as your primary source of news.
Justin Dubs
Re:Enterprise-ready (Score:3)
The fact that you even mention NT as an example of an "enterprise-ready" system shows that you have no idea what you're talking about. NT5 only runs on x86 hardware, which I don't acknowledge as a platform for doing anything but playing Quake 3 or running "myhomepage.com".
"Enterprise-ready" is an E10k with 64 CPUs, 64GB of RAM, a and multi-terabyte SAN. "Enterprise-ready" is an RS/6000 farm. And to me, "enterprise-ready" is real UNIX. Linux lacks scalablity[1] on non-trivial (read: non-x86) hardware and widespread commercial software support. And decent threads! (Ask anyone who's had to port large UNIX software packages to Linux.)
Another problem with Linux is that the historically best distributions (such as Debian) have no corporate accountability, while the commercially popular distributions (such as Red Hat) are buggy-as-hell and dumbed-down.
Don't get me wrong, I use and enjoy Linux... but I'm a UNIX guy at heart and it pains me to see companies such as SCO say that Linux is competitive with UNIX, which IMNSHO it is not. Real UNIX is not only guaranteed to work better than anything else on your vendor's non-trivial hardware, but it will also comply with a large number of open standards that Linux hasn't even approached yet. Each of our ten RS/6000 cages contains about $800,000 of hardware; the cost of UNIX is insignificant, and the advantages are innumerable. The only place for Linux in our datacenter is embedded in our routers.
Offtopic: While I am a UNIX zealot, I like much of RMS's philosophy and use a lot of GNU software. In fact, I refer to my UNIX boxes as "GNU/SunOS": SunOS/Solaris with a full GNU development suite. Incredibly slick. Free Software rules, Open Source drools.
Oh well, no one on this site listens anyway...
[1] And on x86, it only scales well to 4 CPUs, but anyone who is going to waste their money on a (NUM_OF_CPU >= 8) x86 box deserves what he gets.
--
Ellison: How are you gentlemen !! All your database are belong to us
Re:Enterprise-ready (Score:2)
If E-garbage-mart run in a Tandem system they _know_ that if they can't run their business because of a problem with the Tandem servers they know that it'll be fixed in 4 hours, or they get compensation. I believe DEC used to offer _1 hour_ onsite support in some parts of Europe (Germany).
(This means that they needed a dozen sites in the country with people on call 24/7.)
If companies offer cold hard "99.995% uptime" guarantees on Linux systems with compensation terms, then people will consider them Enterprise ready. Not one line of code needs to change - simply that fact that someone is willing to _fiscally_ underwrite it.
FatPhil
--
Re:Enterprise-ready (Score:1)
Also, which "open standards" do "Real Unix" operating systems support that Linux does not?
What makes something a "real unix" anyway? Costing money? Is BSD Unix?
Re:Enterprise-ready (Score:1)
Re:Enterprise-ready (Score:1)
"It will probably take another three years to build a [truly enterprise-ready] Linux kernel."
He seems to be focusing on -just- the kernel, apps be damned.
Re:I can only say... (Score:1)
Besides, recent glibc versions are supposed to be binary-compatible with older 2.x ones, no?
Re:But... what is the point? & UKP (Score:1)
I had a meeting with SCO/Caldera last week and as well as LKP they have UKP in the labs which they 'may' release . i.e. A Unixware kernel Personality to hook on as a module into the Linux kernel .
Kinda interesting, but I agree the 'why?' question sticks out kinda large !
Re:Caldera (Score:1)
Re:3 years? (Score:1)
I don't suppose it means anything that Cray's next supercomputer is going to be an Alpha Linux cluster...
Honestly though, Linux lacks the tools to make this sort of thing a straightforward setup. Configuring Linux from the ground up gets you there, but that's a tedious process that most people don't want to go through. ("What, Linux from Scratch? Ummm...can I have ReliantHA, please?") Also, as of glibc 2.1.3, the basic GNU/Linux system as a whole was not completely POSIX compliant. Some of the missing POSIX features (like atomic file locking, which finally works in glibc 2.2) are actually very good features to have ;)
I don't know what Caldera's claim on Linux threads vs SCO threads is all about. If Linux threads don't actually share the code segment's space (which I doubt) then there could be a performance hit, I suppose. A lot of people insist that the Linux __clone() call (the basis for Linux threading) is not "real" threading, but I fail to see where it differs or takes a significant performance hit.
Kelledin Tane, the Dreaming Minstrel
http://kelledin.tripod.com/scovsms.jpg [tripod.com]
Re:If Caldera is really interested... (Score:1)
1) There's a lot of code in the Unixware kernel that's licensed from third parties and hence can not be opened by Caldera
2) Even if the sources were opened, odds are nobody would use them. This isn't because of any grudges or snobbiness; it's a matter of the Linux kernel and the Unixware kernel being somewhat different at the levels where the opened up sources would benefit (non stop clustering, SMP, etc.)
I doubt that there's much that can be done with the Unixware kernel sources when taking these two items into consideration. It would be nice if the sources were opened, but I doubt it would accomplish much.
Re:I can only say... (Score:1)
???
Re:I can only say... (Score:2)
Re:I can only say... (Score:1)
What are you saying that we don't have those problems anymore? Boy is that wrong. Trust me, I work on the front lines (mid sized pc repair shop that mostly services family computers) and Windows has not improved one iota in the stabiilty department, if anything's it's gotten worse.
Re:Another dead Unix (Score:2)
My good friend Wes [slashdot.org] taught me that you can set the I_WANT_A_BROKEN_PS environment variable in Linux to allow ps-ef. Here's a quick snippet of the Linux ps man page:
Set the I_WANT_A_BROKEN_PS environment variable to force BSD syntax even when options are preceeded by a dash.
--
Big F***ing Deal. (Score:1)
Re:Big F***ing Deal. (Score:1)
Re:But... what is the point? (Score:1)
Think about it. If you're an ISV and you want to port your multi-million $$ generating database system to Linux, but you don't want to give up the power behind your Unixware version, you can port to Linux and run the Linux version on both Linux and Unixware w/ LKP. You only have one port to support and maintain, you get revenue from two markets (does anyone using Linux actually pay for software? I don't know of many).
All of this, just by adding LKP on top of Unixware. What's not to like?
Re:Another dead Unix (Score:2)
You mean closed source or proprietary Unix. Red Hat, Caldera, Mandrake etc are all commercial entities whose Unix-like OS exists for the purposes of making them money.
Re:Like Debian/FreeBSD? (Score:2)
Some Linux users use it for the same reasoin they use closed soruce apps on it: because its the best tool for the job.
Re:Linux needs more game developers (Score:2)
But you can't.
* xdnd is poor, and doesn't often work. Try xdnd from konq FTP to a gnome desktop.
* panel applet imcompatibility
* mime type incompatibility
* doubled learning time from two sets of common controls, completely diffferent visual styles
* Component level incompatibility
* A stack more reasons I can't think of right now
The integration between the two is, frankly, a load of shite. yes there's room for two, but now with the current l;evel of incompatibility. Lackk of consistency hurts the Linux desktop far more than competition is currently enhancing it.
It's a joke, laugh. (Score:1)
Unixware that runs Linux. I guess that would be called??? Lunixware!?!?
I crack myself up. :)
---
Don Rude - AKA - RudeDude
Most of you have missed the point.. (Score:1)
3 years? (Score:1)
If Caldera is really interested... (Score:1)
Re:If Caldera is really interested... (Score:1)
I'll assume this is a legitimate quesiton, and not trolling
We should care because of the reasons Ransom mentioned in the article, however much we might disagree with his perspective. UnixWare is an enterprise level OS and has been for quite some time. It is more mature than Linux and has several features Linux currently lacks. (again, see article). If the source code for UnixWare was made available to the Linux development community, those features could be more readily be implemented in the next version of the Linux kernel. Open-source 101
Re:Enterprise-ready (Score:1)
Another dead Unix (Score:2)
Re:Another dead Unix (Score:1)
Re:Linux needs more game developers (Score:2)
I can see why people would prefer Qt (I do), but I can also see why people who have never done C++ would prefer gtk rather than learning a "new language" - why not give people both (or even more) options?
Re:Another dead Unix (Score:2)
Re:This has been in BSD for a long time (Score:1)
Re:Enterprise-ready (Score:1)
Re:Enterprise-ready (Score:2)
ostiguy
Three Years? (Score:2)
But by then 2.6 will almost be ready for release! That'll be a tough act for Caldera to follow...
--
'ps -ef' in linux (Score:1)
Just get him the right version of 'ps'. I run SuSE 7 and the 'ps' that comes standard with it will do 'ps -ef' just like any good old SysV unix.
Re:Enterprise-ready (Score:1)
Re:I can only say... (Score:1)
Well there's your problem. Of course THEY haven't advanced at all. Win95 was their last advancement in that product line. Win98/ME are just repackages of 95. I'm talking about their NEW OS. Windows 2000.
We run Win2k on a bunch of computers are work and most of my friends run it at home too. I use my for a web/ftp server and as a gaming/dvd machine. It has never crashed. Not even once.
Yes, 2000 DOES still have DLL problems. They aren't NEARLY as bad as they used to be. I was just trying to point out that Windows isn't alone in these problems. Linux has them too. I haven't had any yet with this box though. Mostly those problems are with their home product line.
Justin Dubs
Caldera == Disney? (Score:1)
Anybody else notice that the Slashdot rendering of the Caldera logo looks a lot like half of Mickey Mouse's head?
Whatcha tryin' to tell us, Taco?
Re:I can only say... (Score:1)
Justin Dubs
Re:Another dead Unix (Score:1)
You're right about SCO and DGUX - the only times I saw those two were when they were being replaced by Linux or AIX.
Really, who cares? (Score:1)
This is almost amusing. I mean, I know that since they bought SCO they had to do something with it...but who are they fooling?
Sure, as people have noted, there's still lots of things that Linux could do better. Both it and the
8*86 platform are not the best platform for true high grunt performance on a mission critical task.
Most Linux people know that, albeit they don't think there's anything that will stop those abilities maturing over time.
...but...
Sco's got the same problem! Sure, they used to be the `leading unix on 8086' but that doesn't help them compete with Solaris or Aix. And while they might try and snag the title `unix' for themselves they don't have the marketshare, mindshare or money to seriously challenge IBM or SUN. There's lots of installed SCO systems, but they're not running corporate data-centers.
And, for those things linux does well (at the moment), like providing cheap servers and good desktops, they can't compete with linux.
So, the true message is, "we realise that most of the new apps are linux apps, but we want to keep unixware alive".
(and yeah, I know I should be Caldera unixware instead of SCO...old habits).
Re:I can only say... (Score:1)
You call it and "old" os cause it's based on earlier work. Would that not imply that linux kernel 2.4 is not a new kernel? It's obviously new, in that it just recently came out. But it's obviously old in that it is based on prior work.
Again, semantics.
Justin Dubs
Re:Existing Unixware is bigger than you think (Score:1)
The OS was identified by login as something like "Bell Labs System V UNIX". Which is closely related to SCO UnixWare, but not quite the same thing.
Re:3 years? (Score:1)
Re:3 years? (Score:3)
An enterprise-ready kernel implies one suitable for running banking systems & very large database systems on. It would also imply a featureset much like Digital Tru64 or Solaris, with failover clustering etc.
With the best will in the world, I'm not sure Linux is currently at this stage, and I do not think it will be for a few years either.
Many businesses deloying Linux as web servers, firewalls, or file servers does not make Linux an enterprise-class platform.
Enterprise-ready (Score:2)
Re:I can only say... (Score:1)
If everything switched to the Microsnark standard, every system would be getting blue screens of death from such complex operations as moving the mouse or opening a second application while a first one is still running. Various distros of Linux may have their problems, but I'd rather not go to a M$ desktop standard (no matter how much they plan on making it look like a Mac).
Kierthos
Linux needs more game developers (Score:1)
Looks like Linux is worrying Caldera... (Score:1)
The Linux 2.4 kernel is not everything everyone is claiming it to be and still does not have a lot of the features the Unix kernel has had for years. It will probably take another three years to build a [truly enterprise-ready] Linux kernel.
So, what Caldera is basically saying is, that they need at least 3 years to recoup their investment in SCO... that statement should scare the pointy hair types for at least that long while the tech guys scurry to gather facts on why they should use Linux over SCO.
Re:2.2 (Score:2)
Big companies use Win95 still.. does that make it enterprise-ready?
And just cause it runs print servers or departmental file servers doesn't make it enterprise ready either.
Do banking systems run on it? Real-time transaction processing for stock markets? Critical insturment control for nuclear power plants/other major industries? No....
That's what they mean by 'enterprise-ready' in this context; the ability to stake the data of your whole enterprise on it.
But... what is the point? (Score:1)
Most free software will build on SCO, I'm sure. And SCO has been round long enough to have it's own proprietary market.
OK, I've not seen Quake III for SCO. But, apart from games, what is the point of this? I don't get it.
Now? (Score:1)
Caldera (Score:3)
Installing it and administering it, you really get the feeling that a whole bunch of people were paid a whole bunch of money to spend months and months going over every inch of it, dotting every 'i' and crossing every 't', making sure everything worked together properly and that everything was consistent and uniform and complied with their policies on how things should be laid out. With Redhat, there's always the feeling that it's just a bunch of stuff that they threw together and stuck on a CD.
I use Debian but I would not hesitate to recommend Caldera to someone who wanted to deploy Linux in a business environment (or just for personal use). It's too bad that they aren't more popular than they are (anyone know the market share percentages for the various distros? Not that it matters, it'd just be interesting to see the figures).
Like Debian/FreeBSD? (Score:2)
Or maybe they'll do something stupid and just emulate Linux with Unixware...
- - - - -
Re:Enterprise-ready (Score:1)
- Payroll Processing.
- Databases (no, not mysql)
- Financial Processing
Why not use distributed processing?
- Complicated
- High Latency for transactions
- High licensing costs. (why license software when you can buy hardware?)
- more staff
- Complicated
POSIX shmosix (Score:1)
I mean, cmon! The POSIX "standard" locale charset doesn't even have 8 bits! --mazor
Re:Enterprise-ready (Score:2)
I don't think the issue is how many processors you support, or how much memory can be addressed
Where are the enterprise-level apps?
I would agree than Win2K DC edition is hardly enterprise-class either!
On an important note, there are NO Enterprise-class systems which do not use proprietry hardware: it's the only way companies can guarantee resiliancy and uptime. The intersting point is this: how the hell does Linux achieve this when it's at it's best running on Intel-based kit? I realise it's been ported to pretty much everything, but that's not the same thing as being resilient on every platform it has been ported to!