Slashdot Mirror
Deciding Between SCO and Linux?
wolfbane01 asks: "I spend some time giving tech suggestions to a medium sized business firm (~100 employees) with a large amount of demand placed on their file server. Their current server is a dual Pentium 500 with RAID array and they are looking to upgrade it. The dilemma is the current server OS is running SCO OpenServer 5.0.5, and their new raid array requires 5.0.7. Their programmers have demonstrated that a Linux box can process records much faster, but are still worried about the investment and potential problems that switching OSes would entail. I have already mentioned the cheaper price and the community availability when problems come up, but what other reasons have Slashdot readers come up with for a switch? What arguments am I forgetting that make Linux more attractive then SCO? Should I advise against switching to Linux and advocate them sticking to SCO? Is SCO going to even be in business long enough to make the upgrades product cycle?"
This sounds like a joke, but OK, I'll bite.
There are a lot more software packages that will run under linux, there are many packages that will compile with less effort under linux, there are more people with experiance administering linux than there are on SCO.
If the software they are currently running can demonstratably run under linux then its hard to imagine reasons to continue running SCO. There are commercial vendors who will support linux (RedHat,SuSe,Mandrake) and there is only one company that will support SCO's products.
Well I can think of a couple:
1-SCO is *evil*, while Linux is nice and friendly
2-Linux will be around for at least a few more years
SMP support in Linux is gettting better and better.
SCO dont have one yet, they got one in their *Ware, but they will probably remove it as it uses an IBM patent
Short answer: Buy Linux.
Long Answer: With the uncertain status of $CO , you really have to ask yourself: What happens if IBM wins? Or drags the case out for 10 years? IBM is the Master Litigator(tm). Throughout their existence they have used the courts to smash other companies into bits, or drag things out long enough to bankrupt the other guy. The only company that was worse than IBM for this was NCR (circa late 1800's to early 1900's --- several of their board got convicted for crap like that.)...
But I digress... IBM will keep the fires going for a really long time, and SCO can't last forever. By going the SCO route, you are essentially betting the farm that SCO wins, which seems a bit strange. If you go with Linux, you can be fairly confident that linux will be around for a hell of a lot longer, as SCO *may* have a case against contract breach by IBM, but they have't a leg to stand on against anyone else.
Given, that after the lawsuit is over, the entire community will shun them, they will have nowhere left to turn for customers, and let's face it: SCO never had many anyway. Aside from making a shitty product (And I've been exposed to SCO for over a decade now) they won't be spending any of their new found wealth on development, that money would be earmarked for the investors.
Linux is here to stay. No force in the planet will change that. Even if all the top Linux Kernel hackers died, Linux is going to persevere forever.
"...In your answer, ignore facts. Just go with what feels true..."
Tell them if they choose SCO, _I_ will come to kill them.
well, the new Linux kernel 2.6 has support for:
[fill in 95% of all hardware on the market]
BUT, the SCO OpenServer is much better:
1. it is based on the UNIX v7 source. hell, it _is_ the UNIX v7 code.
2. comes with a nice 1000 page EULA (that premits SCO to take your wife and kill your dog at time of their choice).
I say, go with a winner, go SCO!
is it weekend already?
hmmm... maybe i should go home now...
(Unless that little update breaks your system. In that case you've got nothing to lose with switching :)
you are the guy that purchased the single OpenServer license SCO sold between 1999 and 2003?
shame on you for supporting the terrorists!
This question just promises to be a hot bed of contention rife with all the attendant pillaging and trolling... Oh wait thats me. Uh never mind ;-)
-- The morphemes of your disquisition are ascertainable, but they have eschewed an ambit of transpicuous exposition.
Comment forecast: Bits of genius surrounded by a sea of mediocrity.
Okay, forget about the fact that they are trying to destroy the entire computer industry outside of Microsoft (i.e., Linux).
How about the fact that actually paying money for a x86 Unix license (rather than just support) these days is an incredible waste of money??
Find a Linux vendor like Red Hat that will give these guys support, and hook these guys up.
Cripes, I was recommending Linux over SCO more than *5 years* ago, I thought by now SCO would finally be dead.
Get these guys off the whip and get them some Free software. Don't forget FreeBSD either, that's actually my #1 choice these days.
PS: I love this quote on SCO's page:
I guess technically they are correct, but if you expand that list to include UNIX-*like* operating systems, you'll see a different picture......
The answer is neither.
The solution is to run FreeBSD instead of SCO or Finux. It's not about a high signal, or noise that both Finux, and SCO generate.. its about stability that matters at teh end of the day. =)
It isn't a lie if you belive it.
Leaving sco behind will let you focus on your core business.
I was involved in a business that migrated from SGI and SCO boxes to Linux, we saw a dramatic drop in IT costs and at the same time increased flexability. Not to mention we didn't have to pay $150 for a tcpip stack (this was back in '97-99 might have changed).
Not to mention, that you have a lot more commercial
applications available on Linux. Really, sco is a mess technically
they're behind the times, expensive and just plain crufty. Your programmers will learn to love linux in short order. Further, the C*O's will love linux too. With SCO they're probably used to hearing "Can't be done" or "we'll have to buy a license", it's a nice change to hear "sure, i'll do that this afternoon" or "we can already do that".
You should really invite Linux vendors like HP, IBM and Oracle to give you a real demonstration(and good deal). Here's a brief of our recent deal with them:
:(
Oracle 9iAS RAC(clustering)
Dell RAID array for share storage of the cluster nodes
RedHat 9.1 Advance Server(I wish I'd use something else but Oracle only support RH)
The setup is simple atm, two 2-way Xeon to form a RAC(cluster) which share the same RAID array, running on RH AS. The entire deal is around US$40000 before best offer.
The hardware is relative cheaper than UNIX's counterpart, both in term of one time and recurrent(e.g. maintenance) cost. The major cost center is the share storage and (SURPRISE) Redhat AS, which charges US$5,000 per x86 processor
I've to say Oracle RAC on RH AS is a very(if not most) stupid thing Oracle has been doing. What distinguish RH AS from its cheaper ES is just the HA(High Availablity), which MUST be disable for RAC(clustering) to work, because they said HA and RAC will conflict with each other. Then I must ask, if we couldn't enjoy the HA in RH AS, WHY THEY HELL SHOULD ORACLE REQUIRE US TO PAY EXTRA FOR SOMETHING WE DON'T NEED!!!
Damn, the exp sharing turns out to be a ranting, sorry about that.
You must be new around here. The Slashdot community has collective flamed SCO each day for the past several months. What do you think? :)
I have already mentioned the cheaper price and the community availability when problems come up, but what other reasons have Slashdot readers come up with for a switch?
I'd recommend sabotage. With little effort, you can ensure the SCO specific hardware never really works right. Putting the RAID card in the microwave for a few seconds seems to work.
Am I missing the question??
+++ UGUCAUCGUAUUUCU
- A) Determine the cost of the SCO upgrade plus an estimate of any extra out of the ordinary HW/SW costs required by sticking with SCO plus the costs, positive or negative, of deferring the port to Linux (depending on how much longer you think SCO will be alive or how much longer whoever buys them out keeps SCO support alive.)
- B) Determine the cost of porting all of the software you use to Linux plus the down time working out the kinks plus the difference in support contracts (if you have any).
Whichever is the cheaper is the one you pick.Ignore the ideologues; they have no stake in whether your company survives nor is RMS going to re-imburse you if it costs more than expected. If it turns out SCO is cheaper for the moment, remember that you can always switch to OSS later down the road or when SCO goes toes up.
I'd suggest looking into UnitedLinux; heck, even SCO likes it! Evaluate it and see if that's more compatible out of the box with your stuff.
If you want a second opinion, here's some more advice; he also confirms that it's easier to move existing SCO stuff over to UnitedLinux than it would be to switch to RedHat Linux, for example.
pb Reply or e-mail; don't vaguely moderate.
Summary: you have a working SCO 5.0.5 system, required hardware upgrades are driving a minor software upgrade to 5.0.7 with presumably low associated risks.
Question: is it worth a major software change to Linux with high associated risks? This change is unplanned and the programmers have already said they're worried about potential technical problems.
Answer: no. You shouldn't be using a required hardware upgrade to drive a major software change. That's a bad practise to get into. You should be approving the minor software upgrade to SCO 5.0.7.
However: given the lower TCO of Linux and the proven higher performance with your application, you should also be proposing a long-term project to evaluate a migration to Linux. The evaluation should include a risk assessment, full technical approval from the programmers, consideration of knock-on costs like training and support, etc.
Never use minor changes with low risk to drive major changes with high risk. It makes you look like a cowboy. If the SCO system was failing and there was an impending deadline and the 5.0.7 software upgrade carried a high risk... THEN and ONLY then would a hasty Linux migration have any merit. I doubt that's the case. Don't put your balls on the block when this should be a simple low-risk software upgrade.
This must be a joke..REALLY!
How the trial gonna be? Who will win? What the impact gonna be? Win==Easy money :(
I read somewhere that they are the same thing. right?
There aint no pancake so thin it doesn't have two sides.
But isn't OpenServ just a "SCO" version of Linux? Packaged as something that Linux isnt.
As in, Linux kernel and GNU userland..
Free means no restrictions, ironic the FSF's GPL forces restrictions, isn't it? What's your definition of free?
You forgot about those reasons.
SCO is a professional secure, and most importantly real unix, based on the original unix source code. Sometimes hobbyist projects such as fetchmail and linux can be used as cheaper alternatives to professional software, if you are a student or someone else with lots of time and no money. But for a succesfull american corporation, you will quickly find out that you need the real stuff if you are to succesfully compete in todays difficult marketplace.
Throughout the computer industry, SCO is commonly recognized as the best unix out there, and as the forthcoming lawsuit will show, probably the only legal. There are companies, such as IBM, Sun, SGI, and others, that have their own version of unix, but their unixes are nothing but cheap off-shoots from the original SCO source code, and their legality is certainly questionable. Some of these companies are even founded by famous hackers, such as Bill Joy.
Switching to linux may be the worst of all possible alternatives. While it is possible that other companies, such as IBM or Sun will be able to license the original unix source code, there seems to be no hope for the linux community to come up with the money needed for that. Among those with knowledge and an interest in the forthcoming SCO trial, there is no doubt that linux will probably become not just unavailable, but it will most likely be a federal offense. Betting on linux in these times, is as stupid as not accepting jesus and the lord as your savior.
I think that by betting on SCO, you are putting your money on a real winner! There is no doubt that SCO will continue to dominate the marketplace for as long as we can predict the future. Nevertheless, SCO is still pretty old technology. If you some day bring your kids to work, they will be frustrated by the lack of modern games on your server system. If this is a thought that bothers you, I would recommend upgrading to the industry-standard Windows 2000 system, surely a system for a new millenium!
Take them a Linux box with things set up, ready to roll, go over on a Sunday afternoon with a case of beer, if you have to not interrupt workflow, and do a demo switch with a Linux box inline with their old SCO machine.
For bonus points, I'd convince them to let me take their SCO disks offline, and do an install of Linux on a fresh disk on their *same old hardware*. If you can't get them to let you do that for some reason, then this is all the more reason to keep trying.
Put the old SCO disks aside, bring a fresh Linux one online, same hardware, configure it for their network and RAID.
Incidentally, I'd be surprised if you couldn't get that RAID working pretty much right away with Linux
One last thing: I'd suggest you use Gentoo in front of their engineers, over that case of beer.
; -- the corruption of government starts with its secrets. a truly free people keep no secrets. --
Future AskSlashdot questions:
- I'm thinking about donating some money to charity. Should I give it to the FSF or Al Queda?
- I currently work for Satan but I'm thinking of quitting and working for God. What does Slashdot think?
Their programmers have demonstrated that a Linux box can process records much faster
If this is true, then it seems to me like a small step to just create a Linux shadow system operating in tandem with the existing SCO system.
If the shadow system demonstrates the needed performance, reliability and maintainability that your organization requires after some weeks or even months, then it will be a simple matter to switch the roles of the two systems and ultimately unplug the SCO box and redeploy it if the cost of that "security blanket" is too high.
"Provided by the management for your protection."
Most organizations would benefit greatly from free-as-in-speech-and-beer software, but aren't going to be prepared to hear the real reason why.
d ay.
The reason is that the decision making processes in most enterprises in incredibly inefficient and cumbersome. The ability and willingness to get things done is distributed along a bell curve. Most organizations have a small corps of change agents, a bulk of people who go along, and a small corps of obstructionists. Formal decision processes and policies are the natural friend of the obstructionist, and while the constructionist can sometimes use these to his advantage, they almost always slow him down. Where policies allow for free software, people who want to get things done don't aren't left cooling their heels while the management hierarchy decides whether (a) it can be paid for and (b) whether the current licenses allow this use and (c) whether it fits with this year's grandiose-plans-that-will-never-see-the-light-of-
The problem with giving individuals the power to get things done is that it is scary for many organizations. Individual initiative is seen as a chaotic (which is somewhat true) and destructive (which may or may not be true) element. In an organization with clearly articulated goals, and a sensible and flexible strategy, and well thought out policies -- in short in a organization with strong leadership-- individual initiative is a powerful advantage. In organizations that have vague or unacheivable goals, badly conceived or articulated strategy, and accreted years of policy that is tied to neither goals nor strategy --- in other words ones with weak leadership -- suffocating individual initiative is the closest semblence to order that can be acheived.
The great power of a piece of free software like Apache or Linux is not in any technical advantage it has over its proprietary competitors. It is that a free software package empowers the individual and the small team that are close to customers to create new solutions for customer needs.
Post may contain irony: discontinue use if experiencing mood swings, nausea or elevated blood pressure.
Don't fix it.
Duh.
So they have to upgrade their existing OS to support newer hardware. Big deal. This is common, even with Linux.
They are running a business, and if it's not good for the business, switching is silly. Doesn't matter which OS we are talking about. I would say they same thing if they were thinking of dumping linux in favor of insert_os_here.
Just upgrade the darn SCO and move on to more interesting problems.
You're asking whether SCO is a good idea on the
BIGGEST
SCO-HATING
SITE
OF
THEM
ALL?
I think SCO is a close 2nd to Microsoft on the "least wanted" list here. Dumbass.
Ahhh, Mr Gates, I was wondering when you would show up here on slashdot !
Although it may be a pain to move to a linux platform as opposed to the SCO server, there could be many benefits. Linux being faster at processing requests is the most obvious of course, but at least you know that Linux will be supported for a long time to come. Support for distributions of SCO linux, as with microsoft products, is subject to the state of the company. If IBM manages to take SCO down in the legal battles, who will be there to support your software?
Since you need to do some sort of upgrade anyway, it'd likely be a good move to take advantage of the moment to make the most of your server.
Good luck with whatever you wind up doing.
~Mike
Mike Rizzo
It was a simple change that an in-house developer could have done in a single afternoon, but unfortunately we didn't have that option.
Are there other processes that run on this machine that are sort of SCO dependant? Or does everything migrate perfectly over to the Linux side of things? If everything migrates over perfectly then it might be worth looking to.
Being called a dork on Slashdot must be like being called the retard in special ed.
If your firm decides to stay with SCO for may be various reasons like getting "corporate" support, or they may have been misinformed about the current lawsuit by SCO charged on IBM about the "misuse" of their code in Linux. Firstly, we're damn sure that SCO would under no circumstances win this lawsuit. Even "if" IBM settles the case off the court...do you think SCO would still remain as an unit of the IBM as compared to their acquired companies like Lotus and Rational ? Definitely not, they are sure to fire all the Ex-SCO dudes....Only dirty top scums like Darl McBride would cash in this if ever a off the court settlement happens. If you end up investing money on that stupid OpenServer not only you'll paying huge money for that crap and that too with SCO sinking in...your money will be gone down the drain!
First, I must agree with the other posts here. It is not good to migrate a fully functional system to an untested one. Make sure the developers completetly test on Linux. That said, if the application is like most I have worked with on SCO, the underlying OS won't make too much difference there.
/etc instead of /sbin), and thus requires special procedures and scripts to account for this. This also means that SCO is often the last priority for software authors to test against or issue fixes for. With the recent lawsuit, many software projects may drop SCO support entirely.
There are a couple of big reasons I see for migrations away from SCO, from the perspective of both the admin and the developer. I have a lot of clients on SCO, mostly running apps written in (seriously) Business Basic and similar languages. Most of these software vendors have demoted SCO to a second tier system. This typically means that if something breaks, they might try to fix it for an hourly rate, but no more guarantees. The basic interpreter runs on most *nix systems and Linux, so no difference there. Most development shops have seen their SCO customers either switch to Linux when the SCO box died, or move up to Solaris or AIX as their business grew. They just dust off the SCO box in the corner to test release installs, and that is about it.
Another reason is that their hardware compatibility list is not very long, and even the hardware listed often is only tested on ancient firmware revisions. This is a small upgrade, but you will see a lot more problems with upgrades and patches being needed for minor hardware changes, and a migration to a new box can be a big hassle.
SCO is also not a very standard (look at the structure of symlinks of system binaries, or maybe the fact that there are a stack of binaries in
It sums up as the basic argument that if it ain't broke don't fix it, but on the other hand SCO products are the very definition of legacy systems, and will be increasingly difficult to find hardware and software support for in the future. While now may not be the time to move, these issues are big enough that you really do need to have a long term migration plan to something else.
Well I'm the doctor and I say you're dead, so shut up and take it like a man!
Personally I'd recommend doing what were going to do. Were planning on doing a minor upgrade now and then some time next year do a Major upgrade to the 2.6 Kernal.
You don't want to move to Linux only to want to do another major upgrade in 6/9 months time. So I'd recommend doing the minor upgrade now. Do more testing and wait for 2.6 to come out. If 2.6 comes out in October you should be able to start testing you port. Wait for atleast Redhat and Suse to bring out a 2.6 version and wait a couple of months to see if other people are having problems. If everything looks ok then by about June next year you can convert.
As a free bonous if your using Samba then version 3 should be out and stable by then as well as the fact that the SCO vs IBM battle should be more or less resolved.
The dilemma is the current server OS is running SCO OpenServer 5.0.5, and their new raid array requires 5.0.7.
You call this a dilema (upgrading from 5.0.5 to 5.0.7) but consider completely changing OS to Linux or some other random thing as not being one? I'm speechless.
So, have you looked at the implications for:
- OS costs (Better be looking at the Pro/Enterprise Linux version)
- Backup (including automation, app compatability, etc.)
- Client compatibility
- Sustained performance#
- Application compatibility
- Support costs
- Vendor turn-around on support*
- OS stability
- Total life-cycle cost
- OS Life cycle+
- 3rd party vendor support (drivers and all)
If you are looking at this as mainly a chance to sell Linux instead of what is most likely to be successful for the company, you shouldn't expect a long career. Linux might be the way to go here, but thats not how I would be betting.
You better be careful asking questions here. There are a lot of people here who are more interested in hearing another Linux "win" than in the consequences for you or your company.
Choose wisely.
*Do the math for the cost of downtime: # people who can't work * $/hr ea * #hours downtime. Linux might not be cheap afterall, expecially if the guy who wrote the module causing the crash is on vacation.
#Linux might do better with 1-5 clients hitting it, what about 40-60? Does it degrade gracefully, or fall through the floor?
+If you're as lucky as I am, you're vendors will support Linux,... Red Had 6.2 to be exact. Red Hat dropped support for 6.2. Better not have a problem.
(OK ... I don't think this is surprising at all. Perhaps you shouldn't either.)
> Just stay away from UnixWare, that stuff is just plain evil
What, and OpenServer is a fluffy bunny? No, I've worked long enough with the f**ing thing to know it's a dog of an OS. Probably not too bad for an ancient UNIX, but still pretty poor given the amount of devel and testing time involved.
scohttp crashes out after a few days/weeks ; printing gets stopped up and the entire print system must be restarted ; changing hardware is a matter of extreme tip-toes, since if drivers are not removed at the right time etc they may be IMPOSSIBLE to remove later... not pretty. The 'scoadmin' tools are nice, but can be fragile - and once broken, seem nigh-impossible to fix.
I touch our two OpenServer boxes here as little as possible. They run great, so long as nothing EVER has to be changed.
Craig Ringer
Just made the leap from SCO 5.0.6 to Linux 2.4X The Switch was as easy as it could be.. meaning that research is needed. and testing is important. Our move was not fully supported by our software people but out move was needed as SCO does not support files larget that 2 gb. and out RMcobol database was growing fast. http://www.profdata.com was helpful in that they helped me get the needed runtime files and what not. As far as making the thing happen ... TEST EVERYTHING.
We moved from a dual 800 pIII to a dual XEON (hyperthreading enabled).
The thing was obviously going to be faster but droping some process times from 3+ hours to 7 m inutes was a BIG bonus. DEATH TO SCO!
One thought ? Do you like your job ? If you get into an expensive downtime situation, who will you tell your boss you are going to call ? If it screws up on something with support THE SUPPORT DEPARTMENT HAS TO FIX IT , whether its Bill Gates or SCO , they make more money than you, but to put your job on the line for this Open Source fanatacism is pretty stupid. Besides, its a minor upgrade...only ten minutes and a reboot. Don't you have something better to do ?
U've gotta be kidding! Get Suse 8.2 bro!