Slashdot Mirror
Using Debian in Commercial Environments?
sydb asks: "I am currently persuading my employer to try out Linux. We are heavily dependent on IBM software technologies just now, and it's a very conservative operations organization. As a challenge, I am trying to persuade them to use my preferred distro but there are hurdles: IBM doesn't officially support Debian as a platform, though I have anecdotal evidence that most of it can be persuaded to work (with alien etc). Does Slashdot have experience shoe-horning Debian into this kind of scenario? Most importantly, how have things gone getting IBM support? My rationale for pushing Debian boils down to its vast array of packages available to apt-get, easy upgrades, apt-get itself, and the overall quality and consistency of the system."
Imagine if you tried to introduce them to Gentoo! They'd probably faint.
Despite the fact that my employer has a software environment that they are comfortable with, and that I have very little to gain and everything to lose, I have moved my software evangelism to the workplace. Can you help?
I put Debian on my Thinkpad a20M, and it worked just fine, especially when I put in an eepro ethernet adapter. The windows driver I downloaded from the IBM site for Windows 2000 professional didn't work at all. Debian had the driver right there.
Ask them to read and point to Bruyce Perens previous papers and work.. he was the former head of Debian/GNU and now heads the UserLinux project..
:)
just goolge the name and you will find his website with the paper links..
Or the hard way.. start your own business and demand it as per your ceo status.. I went the hard way
Don't Tread on OpenSource
In general, you're buying IBM software because you can call them up, tell them "it don't work, nosirree" and your contract says they get to send out some engineer(s) and make it work.
If they support your environment.
The gains you might think you'll get by using Debian are absolutely not worth losing your service contract, which you've likely already paid for. There's nothing horribly wrong with SuSE or Redhat, both generally supported IBM environments. If you succeed in getting your boss to install Debian, you're on the process of going up a river without the proverbial paddle.
MORTAR COMBAT!
---- death to all fanatics
You want to put Debian on the systems because of the vast array of software available for it.
They want to run IBM solutions because they can trust that the few apps that they actually want to run on the system will run with no trouble.
The trouble here is that you want Debian on the systems for your own selfish reasons. They want to run their systems as reliably as possible. Since this is a business and not a college dorm room, the business case will always win out.
Debian is a fine distribution. But no company in their right mind would go through a migration just so you can install the latest and greatest software via apt-get. You see, they've already got the software they need running on the system.
I cannot speak about the IBM support, but I can speak about using a less main-stream Linux distro, such as Debian in a serious, commercial software development shop. What I found was that a lot of time was wasted on getting some of the more complex applications to work on it (e.g. Oracle 9i), while getting the same sw to run on something more 'standard', such as RedHat, was a bit easier. In fast-paced environments where every developer's day counts, this does matter. This experience is a bit over 1 year old, so maybe (hopefully!) things have changed since then.
Simpy
you focus on whats best for your company and ultimately your client by using the right tool for the right job instead of trying to hammer the proverbial square peg (Debian) into the round hole? Sorry to not really answer your question but hobbies and personal preference shouldnt take the place of a better solution (e.g. whatever distro of Linux IBM prefers for their hardware)
If I were you, I would find out what distro is acceptable to your Boss, and move to that distro first.
:)
And like others said before, once he's hooked, the rest is history
It's difficult enough as it is to convince PHB switching to Linux, and I wouldn't try jumping over two hurdles at once.
Uselessful technology (Air-Charged
I went through this same discussion at my company, as Debian is my preferred distro as well. The thing is, beyond the distribution scheme, I really don't get to experience the true differences between the distros, as I'm usually running an unstable release anyways.
The link above also documents creating an apt RPM repository - we did this at my company, and to be honest, 99.9% of my gripes with RedHat went away completely.
I'd suggest looking into apt for RPM, it fixes a lot of the problems, and doesn't introduce those posed by a totally new distro on your production boxes.
Q: What do you think about American Culture?
A: I think it's a good idea.
(adapted from Gandhi)
Wouldn't that result in some kind of explosion?
I would say that what you want to to do is set up a technology demo. Put a server together doing a task using debian. The reasoning being that you have expertise in debian, so it reduces cost of the tech demo if you do and support what you know.
When it comes time to decide on an actual rollout they have to make a decision to go with a distro that they know is proven in their environment, or go with what IBM pitches.
But in either case, what you're doing is making the haters defend on two fronts: the vendors pushing for one linux and you pushing for another. With the debate being "which Linux" it stops being "why Linux". It's a win-win.
"Let him go, Ralph. He knows what he's doing." --Otto Mann (simpsons)
I don't think that Debian is the best choice in this situation. I know the whole subject is "Using Debian in Commercial Environments", but that's like asking "How To Fly From New York to London in My Volkswagen?". Debian is a pretty good distro, with a nice and fairly simple way of updating, but it is in fact too conservative for a corporate environment. Ideally, you want to be at the cutting edge of GUIs and the like, but Debian-stable (I can't imagine using Unstable in a business block) normally lags behind a bit; it's still on GNOME 2.4 and KDE 3.0.
So, what do I recommend? Predictable as it sounds, a corporate version of Red Hat like Red Carpet. It comes with groupware (Evolution), a decent browser (Firefox) and more updates than you can shake a stick at via up2date. You can make a profile to mass-install it on a batch of machines, and they guarantee corporate support against copyright lawsuits to the tune of three million dollars.
Even so, it's still cheaper than Windows, and far far harder to get infected with viruses, trojans and spyware. Everything on it but the Red Hat logos is open source, plus you won't get wormed in the process of installing it. You'll be essentially invulnerable to hacking attempts and the like, and will be able to more easily roll out updates than on Windows.
This should provide the kind of reassurance your employers need.
PS: No, I am not an HP employee.
Ok, I love Linux. I use it at work. I work in a really big, international company.
..
Here's my take .
If it's not supported/approved by IBM and you are dealing with IBM then find out what they support and use that.
Why?
Because 1) it's easier, and 2) you want to succeed.
Your job is not to move the organization. Your job is to make your boss look good. IBM is very very talented at making their customers look good at very reasonable prices. You will make your boss look better with IBM's willing help than by trying to fly it yourself.
Apt-get is nice and all, but frankly, support is nicer. If you don't understand that, btw, then you are not experienced enough to be making the decission on what to move forward with. I'm not saying this to be an ass . . . but simply because it's true. Moving them to Linux is smart, but moving them to something the hardware vendor doesn't support is stupid
Everybody get your fire-retardant suits on for the ensuing flamewar...
The core differences between distros are package management, the version of the kernel, and the version of libc. Debian might work fine for what you want it to do, but a subtle problem might occur that you didn't catch during testing, due to a version difference. I've found that shoehorning, as you mentioned, is generally a bad idea. Shoehorn too much, and your feet will hurt.
Given your conservative environment, I think RedHat's Enterprise Linux product line is more appropriate. RedHat can sell you a commercial support contract, and they promise software updates for 5 years. Also, future Linux admins are more likely to be familiar with RedHat, which avoids needing to learn Debian's quirks. Also, IBM or other commercial software (like Oracle) is more likely to be supported on RedHat.
IBM? Support? Ha! My company (a large multinational financial corp) made the mistake of outsourcing all the technologies through IBM. Some of the stuff works, but their websphere Host-On-Demand system for terminal emulation is crap. The support angle of it is absolutely awful. I have a job thanks to their miserable support of anything they don't provide to us at astronomical costs. My team supports everything they don't. Their policy is, "If we didn't provide it then 1) we don't give a damn about it, and 2) we won't even attempt to help you integrate it into the environment we provided for you." Good luck. It took us years of badgering them before they would clear the way for installling Apache on a workstation to provide automatic updates of image processing. And on top of that, they didn't even try to give us a solution - it was just plain no. When we want to do something now, we just do it. Then hell with 'em. If you can wean your company off of their teat, then my hat is not only off to you, but also covered in mustard as I will be happy to eat it.
I also reply below your current threshold.
We're running Debian on several xSeries systems. At first, we were having problems with server lockups. While it turned out to be a problem with the XFS file system, IBM supported us by swapping out just about the entire server.
They won't support the software, but they will support their hardware running it.
I've had the same conversation in my workplace, when deciding upon a distro to standerise on. It was a tossup between SuSE and Debian.
It eventually boiled down to a single point: SuSE had commercial backing from Novell. Debian is purely a community-maintained distro. If we built a server for a customer, and then that customer decided they wanted to buy support for it, the only safe answer was to use SuSE or Redhat... and frankly, none of us (including the management) liked Redhat a whole lot.
At the end of the day, you need to ask yourself a few questions:
1) Are you happy supporting %DISTRO linux?
2) Are your management types going to be happy with it?
3) Are your customers going to be content with it?
4) Is it compatible with commercial packages? (Really important... although you might be able to shoehorn say, Chilisoft onto Debian, do you really wanna do that across a couple of hundred servers, and then end being responsable for manual updates or whatever?)
In all seriousness, go with Red Hat, you won't regret it. They have the best support I've ever had to deal with and their enterprise line is the most consistent, stable, and feature filled distro that I've seen for the enterprise. I use Debian on personal servers, and while it's a great distro, and Debian stable is *extremely* stable, it is not anymore stable then Red Hat. Also, most enterprise applications are geared towards Red Hat. Alien is a nice utility, but sometimes craps out on me. You'll have no trouble finding RPMs of any major application on linux. Also, I love apt-get as much as you do, but yum is great, up2date is nice(although I rarely use it), and apt for rpm is awesome, although I'm not sure what its like on RH's servers, i've only used it on Fedora. Apt-get should not be a major point in your decision considering that once a server is up and running, you should rarely ever have to install or modify many things (other then security update, which RH handles nicely). IBM can't support Debian's repositories anyway because they have no clue what is in them and they have no jurisdiction over their distribution. Just spend the money on a good corporate server and I assure you that you won't regret it. It will also keep the higher ups happy, and if the shit ever hits the fan you can just toss the problem to Red Hat, who are btw very good and very quick at solving damn near any problem in the world.
Regards,
Steve
make a sandbox running Deb on your network to start showing them what it can do. this is what I did at my work, and it worked. Currently CVS and the Build machine are running my Linux distro of choice; Gentoo, for mainly the same reasons you mention.
RHCE's aren't going to do what we can do with *our* distro's, it's more than just LInux to us.
CB
free ipod and free gmail!
So, you need to ask these questions to yourself and your co-workers:
If you have a stable working enviroment, why change?
Is this move going to be cost effective?
Is the distro I use going to be the proper one?
Why am I really using this distro? If you say, because it is the one I use at home, then you need stop this project right in its tracks.
How easy is it to manage this distro in my enviroment. Running "apt-get upgrade" on 500 servers is not do-able.
Is there proper management software out there for my distro/platform of choice?
Does my software I need even run on my distro/platform of choice?
What about support for my software on my distro/platform of choice?
Can I keep my system software in sync across all servers?
Can I easily manage the distro install process?
Can I trim down the install time?
Can I make the install process automated?
These are just the basic questions you need ask. Don't get stuck on one distro. Be flexable and look around. Redhat or Gentoo or something might be better choices.
Linux O Muerte!
I used to work for IBM in the division that developed DB2 for Windows, OS/2, Linux, and various Unicies (but not OS/400 or other "big iron" systems) three years ago, and worked on code for DB2 v6 through to v8.
At that time, our Linux testing was primarily against Red Hat and a few others (from hazy memory, Turbo Linux, Red Flag, and one other I don't recall at the moment). Debian was not tested at all for any of their products. Red Hat was their primary focus, and seemed to be the Linux platform most of the developers ha on their desktop systems (although a lot of the Unix development was actually done through AIX-based systems).
Things may have changed since this time, but I haven't seen any outside evidence of this. Do you really want to try running these applications on platforms and with packages that the original vendor hasn't done any testing with? The IBM products you mention are not cheap -- why risk having them break by running them on an unsupported platform?
If you're a big account, talk to your IBM account rep and tell them you'd like to move to Debian. You'd be suprised how much IBM will do for a big account (or, at least, would do when I was there).
Yaz.
Stick with the supported platforms, son. Dink around with your favorite distro on your own box(es). I've gone one dedicated FreeBSD 5.1 box and one dual boot windows/Debian testing box. I wouldn't think of pushing Debian branded linux in favor of something like RedHat. With RedHat or SuSE you've got a substantial corporation behind you. Not just the distros but the companies who support those platforms as well.
There's plenty of help on the internet at large, but they arent paid to have an answer to you in any amount of time. They don't even have to answer your questions at all. In fact they could simply call you a tart and a fop and go frig yourself or something strange like that instead.
Evangelize Linux, to be sure. But stick with what's supported. You'd rather have IBM or RedHat to point a finger at when it doesn't work rather than sitting on your thumbs and trying to explain to your boss once again why Debian was the superior choice.
Don't turn your companies first encounter with Linux into a science project with your favourite distro. Even if you've heard it can be 'persuaded' to work. You're on a salary as are others, keep it sweet & simple and don't waste your own time, because that creates an impression too.
Go with a flavor of Linux that IBM supports, then later when you're feeling adventurous introduce a Debian box or two. Making the Linux transition any more difficult than it has to be seems utterly pointless, especially inside a conservative organization. Make sure they take the right lessons away from this, not some ambiguous point confused by distro issues.
Muddying the waters with unsupported distro complications is just bad judgement.
The fact that you're talking about "shoehorning" Debian in, using "anecdotal evidence that most of it can be persuaded to work" should answer your question.
This isn't a PHB issue, either. Anyone with a real production system should be scared off by language like that.
Your scenario is a bit vague.
What would Linux be used for? desktop or server room? Debian makes more sense for the latter (stability, consistency and good response time to security issues) than for the former (unease of install, antiquated desktop on Debian Stable, lots of work needed to maintain essentially your own desktop-ready distribution, obvious support issues with IBM, look on the management people face when you tell them your wonderful distro is based on "Debian Unstable", etc).
Maybe you can make the pill easier to swallow if you go to a more commercial version of Linux first, e.g. SuSE or RedHat? This way you only have to clear the first hurdle of making Linux acceptable in your company. It will still come with support contracts, releases, and other things management can cope with. Not to mention that these distros and others have to some extent caught up with Debian, using apt themselves or yum.
If your setup is Linux for the desktop, how much experience do you have with managing more than a handfull of machines and a couple of users under Debian Linux ? Debian currently makes a fine meta-distribution but don't make the mistake of assuming it will be as easy to maintain as your own machine. You'll have to cope with more user demands than just your own and a wider array of hardware.
What do you mean by "doing everything the Debian way"? Are you saying Debian doesn't adhere to the FHS? Or are you just saying that -- while complying with standards relevant to a *nix -- it does things differently than RedHat or SuSe?
If you're simply saying that it does things differently from RedHat, then who says that the way RedHat does things is "the standard"? As for "special config tools", etc, why are Debian's config tools "special Debian config tools", and RedHat's config tools not "special RedHat config tools"?
It seems to me that your either saying that Debian doesn't adhere to standards (such as FHS), which would be a good criticism (even though sometimes standards are wrong), but in which case I'd want some examples; or you're saying that it doesn't do things the "RedHat" way, which is like complaining about it because all of its programs aren't in C:\Program Files.
PS: Personally, I use Gentoo.
social sciences can never use experience to verify their statemen
Doing something like this is just like trying to use Perl or Python (or Java or whatever) in an all-C/C++ shop for the first time. It may be the best solution for the problem you happen to be solving. But if the company doesn't consciously maintain a knowledge base in the "new" technology, any of the new work is essentially dead once the author leaves. Same thing applies to a new OS, a new third-party app, or whatever.
The best technology solutions are maintainable, extendable, and reusable. And the most common error is to overlook maintainability.
And please, windows gimps with no linux experience need not moderate nor reply, because you don't know what you're talking about.
It is true that Debian does not have much commercial support, beyond Progeny and a few others.
However, it is the easiest linux distro to support, hands down. It is far more deterministic, more polite to it's user base, and far easier to support your commercial software on that anything else (provided you do it right). Why debian is not more popular with big houses is a topic up for grabs, but it has more to do with psychology, intertia and plain ignorance than anything else.
and to those who are saying "shut up and go with what's there" I might remind you that the reason they're using linux in the first place is because users (in this case admins) wanted to use it. The demand came before the supply, OK?
I believe Debian is so far superior to the other distros that wide support for it is inevitable. It makes too much sense. I think partly the reason is isn't widely commercially supported is because Debian spent the first years of it's existence more concerned with infrastructural matters than anything else, without much concern for usability. Now that they are very actively working on usability issues and other assorted superficialities, look out. they have a solid, modular architecture supported by well designed political process.
Lastly I might add Debian is not a company that can be bought or influenced by money; it is a non-profit with protected legal status. It is very politically stable and is the only software producing organization I know of that has a social contract with it users. Gentoo or FreeBSD (both being somewhat "cathedral like" in their organization) may have the quality of Debian, but they can't match the political stability, and neither can any commercial company.
Run your critical application on Debian. But make sure that if and when something goes wrong, you can reproduce the problem on a supported distro (Redhat, I guess).
We have had a government contract that required Oracle 8i for odd reasons. Debian still has available the older libc versions needed for Oracle 8i. I don't if any current versions of RHEL or SLES support 8i, but I know Debian + the older 1.1.8 JDK allowed the Oracle installer to run and work with minimal shoe-horning.
The other Debian box we built for this application was for running Tomcat with the Sun JDK pushing a web-based reporting tool. We were able to demonstrate how Debian supported removing all unrelated packages (including compilers) and lowered the security profile lower than their Solaris boxes. (They still used telnet, God help them) The demonstration worked and the server is running Debian in production on the [redacted] government network.
Don't push it. We recommend Debian because of access to the build/distribution system and the ability to craft custom loads for specific purposes (point-of-sale, thin client, rich client, etc.). Controlling the build/distribution environment is a bigger issue than many people realize. But we really support anything because after a certain point, Linux is just Linux.
Comment on DebianPlanet about how we do it
We use it in our business and support it for our customers. No problems here! Go Debian!
My God! It's full of Voids!
The reasons you cite for using Debian are contradicted by your stated approach. The reason that apt-get and dpkg are strong is the features they have vs rpm--using alien squanders any advantage you may have gained by going with .deb .
You find yourself using terms like shoe-horn, this should be an indication to you that the shoe doesn't fit.
I, too, am in a heavy IBM Websphere and DB2 environment and when we bought new hardware I looked into upgrading the distro from Red Hat 7.3.
First, the install on Debian isn't smooth. I tried the latest stable Debian as well as some updated packages that I knew I'd need. I installed Websphere and had some problems. Stuff worked, eventually, but it was a pain that I wasn't willing to deal with on an ongoing basis (fixpacks and such). Java GUIs were particularly troublesome, although the web console is really all you ever need. Java problems worried me a lot.
I tried Suse and Red Hat's enterprise offerings, which I had been given demo disks for, as well as their free counterparts. One major hurdle with Red Hat was that there are some major Java threading issues with RHEL 3.0 and Red Hat 9 and above, so I'd be stuck with RHEL 2.1 or RH 8. I decided to go with Suse 8.2, which is supported as a development platform (no free Linux is supported for production use).
What I found on my distro adventures is that IBM supports anything, but they do complain about it. For instance, even our old environment had RH 7.3 while only 7.2 is supported. During my Debian install it was IBM who helped me get it working. When supporting these distros they constantly question the Java version and go through a checklist of software versions to make sure everything's ok. But like I said, they will support it.
While I have gotten bad support from IBM before, overall they are much better than any other company I've had to deal with on an ongoing basis. They really do try to help out. A couple times I've had some idiot at their help desk so I asked to be transfered to someone else, but other than that they've been great.
The global economy is a great thing until you feel it locally.
In that case, it is not correct to say that Debian is a "non-standard distribution". It simply isn't supported by Oracle and IBM. This really shouldn't be much of a problem if the distros supported by Oracle and IBM and Debian adhere strongly to standards (like FHS).
As noted in the above message, I don't use Debian, but Gentoo (and I probably wouldn't recommend Gentoo to a corporation, due to lack of big-company support, unless there were special circumstances that hyperbolized the benefits of Gentoo).
I'm not "defensive about my operating system". I'm just curious by what the person meant when they were talking about "non-standard". RedHat is not a "standard", nor is Gentoo, or any other distribution. They are simply implementations. It is simply one among many distributions of GNU/Linux. FHS, on the other hand, is a standard. Thus, any Linux that doesn't adhere to FHS (such as GoboLinux) is non-standard.
social sciences can never use experience to verify their statemen
Many of us inside IBM would like to see at least one free distribution supported. However, IBM won't support Debian unless there's customer demand. You're a customer, so demand it. Keep demanding it.
GCHQ Quantum Insert installed. If only our tongues were made of glass, how much more careful we would be when we speak
You know the right answer. The fact that you're even asking here means you already know deep down that the best thing to do is RedHat or SuSE.
With that said, use SuSE. The last thing we need is more RedHat customers. Competition is vital to keep Linux from turning into a RedHat-only proposition (in the enterprise). Support SuSE, at least keep it a duopoly between Novell and RedHat - they'll beat each other up and keep things fair.
I've picked Debian for an embedded systems project we're working on.
The problem with the distros out there is that some are updated 2-3 times per year (stable release to stable release) and then the old releases are supported for maybe another year.
We wanted something with a longer release cycle. Sure, we could have picked RHEL, but the client is cheap, and didn't want to pay big bucks for support.
So we're going with Debian Sarge. It should go stable well before the project has finished development, and with any luck (i.e. Debian again takes forever to push out another release), we'll still get security updates for 3 years or so.
But this is an embedded application without a lot of external software dependancies. We're using a free database, for example.
I've experimented with several distros, but I've stuck with Debian for our servers and workstations. Our main fileserver, for example, has never, never ever crashed. I'd have 4 years of continuous uptime if it wasn't for various office moves and OS upgrades. I attribute it to very solid, somewhat expensive hardware, a good UPS, and Debian. It first ran 2.2, and now runs 3.0. And in a few months, it will probably run sarge (3.1)... or be retired because it really doesn't have that much free disk space left. :-)
It's truely free and fully open source, support is just about as good as Red Hat or Suse [again unless you're willing to paybig bucks], forward and backward releases are supported fully...no pressure to upgrade on a company's timetable, and software compatiblity is of the highest level... In a nutshell Debian IS Linux!
What's needed in the general OSS movement is to get more corperate interest in the grassroots OSS movements... Personally, I'm a Suse fan...because they have some great IBM hardware ports [like iSeries/AS400!] but realistically, distros like Gentoo and Debian are the future of software...companies like RH & Suse are attempts to strap "traditonal" lock-in software business to OSS/Linux... they are bound to fail...and leave you holding the bag. The beauty of Gentoo and Debian is that anybody can bolt anything they want on to the very stable bases...and when the base changes it's easy to work the changes into your custom software...they are DESIGNED to do just what most companies need!!!
As far as stability and compatibility, isn't it an open joke that the current version of Debian Stable is pushing 3 years old...I'd call that a pretty reliable standard base...better than ANY of the corperate Linuii.
As soon as you have a problem, you get the following conflicting and impossible solutions:
Debian Philosophy says: "Just recompile your app from source"
Commercial interests says: "Just use a supported distribution for our application"
The best thing you can do is keep the Debian box all stuff that complies with the Debian Free Software Guidelines (DFSG) and you'll be fine. If you need something that's no in Stable or not a late enough version in Stable, check out http://backports.org for expanded/updated packages. My last job used an old dual proc P3 running Woody to host our development "all-in-wonder" box - CVS, Bugzilla, CVSZilla, Wikki, development intranet web pages and some supporting tools. We used an rsync via ssh to a Solaris box w/ tape for nightly backups. It worked like a champ for a small team (4 devs, 1 manager & an occasional tester) without blinking. I'm sure would have scaled up at least 5 times that before the hardware we were running it on became the bottleneck.
*** Sigs are a stupid waste of bandwidth.
Why try for Debian? You will fail. But you **MAY** win with Red Hat, and then move to Debian LATER.
"Who are in control, they are not in control of anything - they don't even control themselves!" - Glen Beck
Automatic upgrades from an uncontrolled source are the last thing you want to do in a production environment. Set a standard image, then when updates come along, evaluate them in a test environment, then distribute exactly those updates to the production systems.
Stick to standards, and things you can duplicate exactly, or you're asking for a world of trouble.
-Todd
"The details of my life are quite inconsequential..."
They test new packages and software to death before including it into the official version. The current official version of Debian is Woody, and it uses version 2.2 the Linux kernel. I mean really, you don't get more conservative than that. There is something to be said for using older well tested software. Debian is such a solid founation, it is the basis for many other distributions such as Knoppix, Libranet, Xandros etc.
Comparing Debian to Mandrake, Suse, Slackware or even RHEL I think you will find that Debian it the most cautious about adopting new versions of core libraries, graphics system or the kernel.
I work in a Windows shop. Well, okay, we have a whole IBM AIX side of the company that runs the Peoplesoft stuff, but for all the rest of the company it's Windows. We tie peoplesoft and pretty much everything else you can think of into Active Directory. It works.
But there's places where I can see Linux boxes excelling where other software falls short. One of them is our Spam "solution." It was very expensive and it doesn't work for shit. 80% accuracy, maybe. Lots of false positives. In 2002, it was really cool shit. But that's the problem - things change fast when it comes to certain things like Spam and when you pay $50,000 for a license to filter spam you don't want to upgrade or change softwares every six months.
Enter OSS - My (*gasp*) spamassassin+dspam+amavisd-new is easily doing 99.99% of the spam with extremely low occurances of false positives. Is it supported? Nope. Wait, yes it is. I SUPPORT IT.
Some companies are all about support, support, support. They don't trust their IT staff, they consider them expendable. I don't work at a company like that. They put weight in our abilities. If you can make a good case for an OSS solution, one where you can support it yourself and train others, it will be seriously considered. Apparently there's other companies like this too, since a lot of places are running Linux now and not all of them use RedHat Enterprise.
- It's not the Macs I hate. It's Digg users. -
I love Debian, and I think Debian's package system beats the other Linux systems, Windows, and Macintosh hands down for software installs and maintenance.
But you are dealing with an organization with lots of people who are used to doing things one way, and it will take them time to learn. If you want to convert them over to Linux (and there are lots of good reasons for doing so, including cost and security), pick a distro with a feel as close to Windows as possible. I think (for better or for worse) SuSE meets that goal. RedHat is probably also pretty good in that regard. Both also have commercial support and companies behind them, which makes management happy (even if you don't actually need it).
Change organizations gradually, otherwise you will have a revolution on your hands.
In years past, I introduced more than a few people to linux when we brought in cvs for source code control, bugzilla for issue tracking, apache for our intranet, etc.
Some of these apps can be run on Windows, but we got it running much easier on linux.
If you bring the right (read USEFULL) applications in, linux will sail right in.. because you weren't bringing in linux, you were solving a problem.
Agile Artisans
If something goes wrong, make sure you can blame someone else.
Why do so many people stay with Microsoft? Here is your answer.
If you get a chance to talk to anyone from IBM, make it clear that you'd really like Debian support. Then use a supported distro. Really, this is the best advice you're going to get.
I like and use Debian on all of my computers, including my company-provided T40 laptop. I do it because I like it and because I'm willing to put in the extra time it takes to make it all work. And it does all work, including DB/2 and Websphere and Lotus Notes and bunches of other stuff.
But I still wouldn't recommend it.
Why do I do it then? When I started using Linux on my laptop (my primary workstation), the only officially-supported desktop operating system in IBM was Windows 95. Given that there was no official IBM Linux distro, I picked what I liked, and I struggled through all of the issues to make it work. I stick with Debian because (a) I like it and (b) it's not clear that migrating to the internal (Red Hat) distro would save me any time, 'cause my system works great.
However, if I had to install a new Linux image for work right now (instead of just migrating my old Debian image), I'd go with the standard build, mainly so that I'd get support, and so that every non-Free app I have to install wouldn't be such a pain. I've always run unsupported desktops ever since I worked at IBM -- the OS/2 load they gave me when I started back in 1997 lasted two days -- but it has of late become more and more painful in direct proportion to the amount of internal Linux support, ironically enough.
So my current opinion is that if you're running commercial software on production systems, you should use a supported distro, which means Red Hat or SuSe, pretty much -- and not just with IBM software. Those are the platforms that are supported by all the vendors of commercial Linux software.
Note to ACs: I usually delete AC replies without reading them. If you want to talk to me, log in.
We're in a similar situation but the key here is support. IBM will NOT help you if you aren't running a supported distro. Try running an X445 with ql2300 cards in HA mode talking to a FastT SAN running DB2 with LifeKeeper for failover support.
Now contact all parties involved and tell them you need support. Oh yeah, my distro is Debian. Everyone from IBM hardware to IBM Software to SteelEye will tell you to go suck rocks and come back with a supported distro.
When we did our TSM install, we had an issue with RedHat 2.1 and the 3582 Tape Library Driver. We called IBM and they provided a driver but it only worked on RedHat 3.
What did we do? We upgraded the box. What good is our nice shiny infrastructure if there's no backup?
Now everyone will bitch and moan that you shouldn't lock yourself in like this or that you should just run whatever distro you want. We designed everything about our enterprise app to be portable. If we get tired of Websphere, we move to Tomcat which is our development platform anyway. If we get tired of DB2, we move to Oracle or Postgres or some other database. We aren't using any DB2 SQL.
But until that time, I like the fact that I can make one call and get the support I need. It's IBM hardware running IBM software. The only non-IBM stuff is the OS and SteelEye LifeKeeper. IBM actually worked with SteelEye for us on a DB2 issue with our SAN.
Having said all that, we do use a few unsupported configurations. Our app uses CUPS for server-side printing. Those boxes are Gentoo. Our datawarehouse is mysql running on Gentoo. The interesting part is that I've actually gone unsupported in one area and that's the warehouse. I had to do a bit of engineering to get Gentoo and my two Fiber Cards to recognize the SAN properly. That and I did a custom ebuild of TSM for backup purposes.
All of this leads me to say one thing, if you value your job, stay supported and keep distro zealotry out of the way. If the company is willing to spend on IBM hardware and software then the cost of a SELS or RHAS license is nothing. It will pay off the first time you call DB2 or WAS support about an issue that, while not having ANYTHING to do with the underlying OS (other than it's Linux), they won't help you because you decided to go unsupported. Explain that to your boss as you're being escorted out the door.
"Fighting the underpants gnomes since 1998!" "Bruce Schneier knows the state of schroedinger's cat"
I agree. Debian is wonderful, I use it at home, I use it at work. If your work is expecting to get Enterprise level support, you can get Enterprise level support for Debian with HP.
However, it sounds like your Enterprise has already standardized around IBM. As good as Debian is, I can't see how it's good enough to lose an enterprise support agreement, even if it's just a few machines.
Maybe you can threaten the sales people to go to HP if they don't amend the support contract to include Debian. They probably will know you're bluffing, but it might help.
----
Open mind, insert foot.
Okay...so maybe apt-rpm did not handle package blah. Are you installing blah at work? By your own explanation of what the box will do, this is unlikely, so why do you care? You seem to e predicating your argument on features you will by your own admission never use.
Nope. Point out the very most important fact: Debian Can Not go out of buisness, change its buisness model or decide it wants to charge you 1K for your next upgrade. It also has about 1.5 - 2 years between releases so you don't have to constantly play catchup. It is PERFECT for a conservative environment.
You are wasting your time because you are deploying rational argument in a management environment. These are incompatible concepts.
Managers do not reach decision-making levels in large organisations by listening to rational arguments. They get there through a host of means including but not limited to back-covering, buck-passing, palm-greasing and politics-playing.
Whenever presented with an argument their first reaction will be to do the updside/downside caculation, which goes roughly "If my name is on this and it goes right, how much kudos do I get vs if it goes wrong, do I get the blame"? Nobody to whom the blame sticks progresses up the pole.
If they choose GNU/Linux and it goes right, there will be some bottom-line benefit. A million people will claim that the small bottom-line effect was not due to the choice of GNU/Linux but better outsourcing, maintenance contracts, management or whatever it was that THEY are responsible for.
If there is a single significant failure, EVERYONE will point to the hapless decision-maker and say "See, told you so, this free software is crap and there is the idiot who selected it, no wonder we fsked-up / lost money / had downtime.
Now, put yourself in the position of the person you are arguing with. You are pressing the wrong buttons.
IF however you can pass the blame AND save money, there is a slim chance of getting the argument through, but trust me, the argument will revolve 80% around blame and 20% around cheaper/better/whatever.
Who is your principal software maintenance company (can't see the parent post whilst replying). Was it IBM? IF you can get them to guarantee to support the software and carry the blame, you have solved one of the blockers. Problem is, they only support RedHat and SuSE/Novell.
Unsupported Debian? Forget it, it is a waste of time. Appeals to rationality, quality, goodness of fit are not the issue. Should some remarkable turnaround occur and (say) EDS suddenly announce support for Debian you have the slimmest of chances, but if EDS or whomever aren't already involved in big contracts in your outfit, the supply-chain people will find reasons not to start negotiations with them (risk again) instead of sticking with the existing known supplier (much less 'risk').
Your only hope is to find a friend in the maintainer/supplier and convince those people first. Then they take your manager person out to dinner/golf and start telling him/her why Linux is so good for your business and that might stand a chance of winning.
You don't like what I have said above? Your choice, but I do this stuff for a living. I know the reality.
thousand servers.
My small team got the opportunity to proof a linux desktop system could actually
be useful for the enterprise. The parameters were absolutely ideal:
* Several hundred desktops should be migrated from some kind of arcane unix
to a linux desktop (-> Users who know there is a world beyond Windows)
* All they need is a huge inhouse X-application, that would cost millions to
port to Windows, but is simple to recompile on Linux.
Obviously this situation is a winner to attempt to bring linux to the desktop.
It seems management had heard names like RedHat or SuSE before, but the
they did not have a straight preference. Therefore we managed
to the task without the suits trying to sell their
distro, so we (the techs)
could work with our back free.
Now our big enterprise has a deep integration of windows machines in
its custom environment. We felt the possibility to adopt Debian and integrate
it just as deep into this environment. What does this mean?
* There is a configuration management database with html-interface. We use it to
feed parameters in the debian package management and to configure our systems.
* There is an optimized process to install a custom windows desktops. We took the
process and made it the base for our debian install using FAI and Knoppix
hardware detection. (And yes, we install a system in 15minutes automatically,
while the windows desktops need user interaction and takes
two hours in a lab)
* Our users reside in the Microsoft ActiveDirectory. We used the vintela software
to hook up on the AD as well. This was a decision in order to save some time.
You can do it with non-commercial software as well, but vintela is ready-made
and easy to alienate into a debian package.
* Our enterprise is of course very conservative when adopting new software. So
we took the idea of unstable, testing and stable debian distributions and
extended it by a distro called pilot. This one is very close to the stable one
and basically the end-users getting the software a few days ahead, giving
us another layer of confidence in the stability.
* Our management is very fond of packages. We took this to the extreme as we
saw it suiteable for desktops: Everything is a package now. The root-password
is in a package, the desktop menu is in a package, the special fonts are
in a package and the sources list is in a package. Some of the packages ask
the config database mentioned above for parameters.
All the config packages use cfengine to manipulate the system, if problems
arise we have port 22 open and a service user (distributed as package) who
can be used to install/deinstall packages or to gain root access.
* We did it all within something around 40 man-days. This sounds very
convincing to the management, as they know how many years they paid to
squeeze the same functionality out of Windows.
During our proof of concept we saw it is very simple to integrate all these great
components into our corporate environment. If RedHat would have sold us
their commercial configuration and provisioning module, it would have been their
process and no longer our own well-tested installation and configuration process.
We have a good command-line interface to our
package management and no silly "advanced web
interface" to our package server. It is all
scriptable and we know the scripts as we have
written them.
So under the line debian proofs to be successful here, because it is so flexible
and because we do not need no certified OS to run commercial software. There is none.
What's missing?
You may have guessed: Management is willing to follow our
proposal but they want a support contract with a serious company with few letters in
the name.
We think this is no
I have to add that it was an IBM reseller (but who does the support and everything) and not directly IBM, but Debian was - as far as I understood - their first choice.
conference website Bedfont Lakes is next door to Heathrow airport. my talk is on IBM Collaborative Technologies using Linux. Basically I am demoing Notes under Wine and ways of getting Sametime and ICT going under Linux. For the demo I am using Knoppix installed to disk, which is basically Debian but you know it is going to work before you install it. If you have any questions or if you want to attend then go see the conference website or contact me through my blog.
Either you live in some alternate universe in which vendors work on bugs for individual users, or you've been smoking some exceptionally strong weed. Or, possibly, you don't have a clue.
I don't believe in alternate universes.
I've had this problem with IBM.
I work in Italy. A company that produces an accounting package was interested in bundling their solution with our Debian-based server product.
Their solution uses DB2 for its database. It was important to them and their clients that IBM supported the DB2 installs back-ending their software. IBM only certifies DB2 installs (at least in Italy) on RedHat 7.X and a flavour of SUSE I don't recall now... Yes, in 2004 they will insist upon RedHat 7.X if you want IBM's support. Yes, I pointed out that RedHat doesn't support 7.X any more so essentially they were asking their clients to choose a lack of support for their DB or lack of support for their OS.
I'm sure there are countless examples where heavy-hitting software vendors have been able to cajole support from IBM for other distros but small software companies haven't got a hope.
In a last-gasp effort, I adapted the IBM installation and update scripts to use alien and dpkg and demonstrated that they worked flawlessly. The accounting package developers were happy, we were happy... IBM refused to budge.
We used to use Slackware on our colo servers; but following an incident, which required a re-install of one machine, we were forced to make a choice between SuSE (which we'd have had to pay for) or Debian (which I already knew intimately). Easy decision (and made me less replaceable into the bargain)! I soon had my boss -- an old-skool Unix guru and Slackware devotee -- converted to the wonders of apt-get. (Till a package he wanted wasn't in apt, then he was back to cursing and decrying package management systems of all flavours. But this si normal.)
..... at home, not at work, and make sure you don't have any sharp objects within easy reach .....
The colo machines are running Woody (stable), but in the office, I'm running Sarge (testing) and Sid (unstable) on my desktop, just because it includes the latest KDE. Usual story: needed just one package; tried backports, hit snags; decided what the hey. No problems as yet. Remember, Debian is always more stable than Fedora -- and packages won't get updated unless people actively test out the newest versions and give decent feedback. Also, in Debianese, "unstable" refers not to the behaviour of the software, but to the level of development activity. If you want a really unstable operating system from Debian, try experimental
To summarise, I recommend: Stable for remote servers; Testing for servers you can physically get to and other people's desktops to which you can get root access; and Unstable for your own desktop.
Je fume. Tu fumes. Nous fûmes!
No matter what OS or Application you use, your supposed to TEST everything before you roll it into production.
If your the kind of person who does things haphazard then your asking for trouble. Debian won't make you a better sysadmin.
The OS is only as secure and stable as the person managing it.
I agree with using a supported platform as stated in many comments throughout this post. Since you're not running a server farm, you're not going to really benefit from the use of apt-get for updates.
With SUSE, you can run YaST remotely from a terminal window and perform your on-line updates. You can choose from doing them manually or automatically.
Also, considering IBM put $50 million in the Novell purchase of SUSE, it may even be the safest bet for a supported platform
We're mostly developers, which is probably what made us attracted to debian in the first place. We have a developer in our group that wears the sysadmin hat (ducks) but he is both a black-belt problem solver and a good admin. I enjoy the anal-retentiveness of debian-devel and its great to see so many minds focusing on a project.
We put a lot of faith into Debian. Our servers run all of our models and our execution platform, which trades enough securities every day to put my face on MSNBC if something goes horribly wrong.
We do use 3rd party libraries in our software development, and as far as they know, we're running Redhat like we're supposed to. I have yet to have a conversation with someone in tech support that is really a Linux guru. I'm not going to claim to be one, either; however, the code I support is only used by my group. The people I usually talk to in support are usually developers, too. If our group had to support 3rd party executables, then Debian probably wouldn't work so smoothly.
All these negative comments about Debian have suprised me a little bit. Perhaps I don't read /. often enough. And no, I probably wouldn't recommend Debian to any of my peers outside my company. But I don't think "Using Debian in Commercial Environments?" is a ridiculous question, either. It can work without a headache for a troop of coder monkeys writing in-house software.
It also has about 1.5 - 2 years between releases so you don't have to constantly play catchup.
I think your numbers are low. The current consensus seems to be that the old version of stable will be supported for one year after a new version of stable is released. If the release cycle stays the same, it's more like 3 to 4 years total.
...sounds to me like he's using the standard approach to building job-security...
I'd say all major distributions are about the same.
The thing that keeps Red Hat and SuSE on the top is certifications, validations and things like that, which aren't free.
Is Debian enterprise ready? Yes.
Would I recommend it to enterprise customers? No.
First, very few application vendors explicitly support it. I've had bad experience with Red Hat 8 (a vendor who "supported it" until we run into a RH8-specific bug they couldn't fix, then they recommended RH Enterprise Server) so I would be very very careful about that. This has nothing to do with "skillz" - sometimes to make things work you'd need to change the application or do something which isn't possible.
Second, if you happen to need to connect it to SAN or such hardware (or install Oracle on it), you'll be in big trouble - not because it can't be installed (it can) but because the customer would kill if they knew their 100K of h/w or s/w has been rendered unsupported because you've used an unsupported OS.
Third, in many situations, OS cost is about 1% of total TCO, so why bother?
Debian needs certifications and h/w vendor support. I hope some big Linux user will donate this money to Debian to get couple of important certifications for enterprise h/w and software.