Changing a Windows Network to Linux?

Charli125 asks: "I have been a Windows administrator ever since I got into the tech industry. I have no working experience with any other OS except for those from Apple. I am currently working with a small company that has 1 server and around 15 workstations. I would like to faze out all of the Windows software, in order to avoid having to deal with viruses, save money, and learn how the other half lives. Since I have never so much as installed Linux, I am looking for resources or advice on how to get started, and how to plan the change over. Can anyone help me?" "My current setup is like this:
Server-Windows Small Business Server 2003, Symantec Antivirus Corporate Edition(and for Exchange)

Workstations-Windows XP Pro, Office 2003 or XP, Symantec Antivirus, IE for most of the browsers (Firefox for mine and a few others). The main apps used are like every other office, Word and Excel.

I have OpenOffice installed on my home computer, and it works fine. I do see a problem with non-tech types saving in the default format though, which would result in other users not being able to read the documents/spreadsheets.

Yeah, I know, I'm so MSFT, I'm everything you guys hate about the IT world, but I would really appreciate anything that would help me get started.

Thanks"

Comments

A basic strategy:

[TripMaster Monkey]

Since I have never so much as installed Linux, I am looking for resources or advice on how to get started, and how to plan the change over. Can anyone help me?

Not to sound condescending here, but honestly...if you've never even installed Linux, much less used it long-term, you really have no idea what you're letting yourself in for.

My advice would be to install Linux on your home system in a dual-boot configuration with your existing Windows install. Then, use the Linux system exclusively. If something comes up that you don't know how to do in Linux, don't boot into Windows...figure out how to do it on Linux. Only switch to Windows if you have no other option, and if you do, keep a detailed record of the transgression in a notebook (hereafter referred to as the 'Book of Shame'). Also, remember to try to do things your users at work will be expected to be able to do...after all, you're evaluating the feasibility of Linux in your workplace as well as your home. After you've figured out how to do something you couldn't do previously, check off its entry in the Book of Shame.

After a few weeks (months?) of this, sit down with your Book of Shame and see what you've learned. How many entries do you have? How many were subsequently checked off? At this point, you should have a good idea of whether or not deploying Linux in the workplace is a viable option for you.

One more thing...use the resources available to you. Here's a good link to some online Linux courses...you might want to check them out for starters. Also, the Web and the newsgroups usually have good information about whatever question you might have, if you have the patience to dig deeply enough, and the self-esteem to withstand the few elitist jerks who scoff and call you a n00b. Finally, read those man pages!

I'm looking forward to seeing what you find out...please remember to let us know. ^_^

Re:A basic strategy:

[mysticgoat]

In addition to Parent's good advice, I'd suggest looking at the Munich strategy for Win -> Linux conversion. As I understand it, they are basically moving a lot of their folks from MS Office and MSIE to FOSS equivalents, but holding off on changing the OS itself until employees are up to speed with the applications. I think that would be a good strategy for a lot businesses.

Re:A basic strategy:

[geminidomino]

Good advice, except with one point I disagree with.

DON'T dual-boot. Makes it WAY too easy to use windows as a crutch. When I finally went pure-linux on the desktop, the only way I managed to pull it off was by purging the MS software altogether, so I had no choice but to make Slack work.

Faze?

[Anonymous Coward]

I would like to faze out all of the Windows software
Answer.com definition of faze is interesting in this context -

faze (fâz) pronunciation tr.v., fazed, fazing, fazes. To disrupt the composure of; disconcert. See synonyms at embarrass.
If that's what the poster wanted to do, isn't he wasting everyone's time? (I think disruption comes free when you have Windows servers and workstations - or did MS start charging for it separately?)

Oh dear!

[Anonymous Coward]

"I have been a Windows administrator ever since I got into the tech industry. I have no working experience with any other OS except for those from Apple."

Poor customer. Seriously, I've got a Linux and no Windows background, but wouldn't be ready to dive into such a venture by asking advice on some random website. For god's sake this is Slashdot!
You must be new here.

No offense, but

[Phleg]

Don't convert your shop to Linux. Especially if you've never so much as installed it before. You will run into problems, and you will have no clue how to fix them, pissing off your users and disrupting your company's business. No matter how bad it is running Windows, it will be worse running any platform you aren't familiar with.

Install a system for you home use and use it exclusively for as long as it takes to get comfortable. Set up some services on those machines and tinker around with them. You should have, in my experience, at least a year with the operating system before you even think of deploying it in a business environment. Less, if you're familiar with similar operating systems.

I don't buy it.

[Seumas]

How this got accepted on Slashdot, I can't imagine. It's clearly not a sincere query.

First, he says his only professional experience has been working with Windows machines. Then he says that the only experience he has is with Apple systems. Which is it?

Second, he says he doesn't know where to start with regards to involving linux with the network, because he's never so much as installed linux before. Well, the obvious first step would be to start by installing linux.

Third, he ends with a comment intended to do nothing but rile up the Slashdot crowd, because he's "such a MSFT boy". Which doesn't even make sense since in his first sentences, he stated that his only experience is with Windows, but he only knows Apple.

Seriously. Come on.

Only slightly kidding here...

[evilpenguin]

Hire an experienced Linux administrator.

Seriously.

The first post on this thread has a good suggestion on how to begin getting personal experience with Linux, but seriously, you are talking about pretty big change here, even if you only have a few servers.

Which distro will you choose? Why?
Do you know enough about the system to handle a problem when it occurs?
How much does downtime cost your business?
Would signifcant downtime due to your ignorance of the platform cost more than maintaining and securing the Windows solution you already have?

If you don't have the expertise, it isn't reasonable to ask your employer to pay for you to acquire it in production.

Believe me, I am all for converting businesses, especially small to medium sized businesses, to an open-source infrastructure, but at first glance this seems like you are just begging for trouble.

I'm quite serious when I say if you want to do this, contract out to an expert who does this. I charge $95/hour plus travel/lodging expenses... ;-)

Otherwise, maintain what you have and develop sufficient expertise on your own. Convert when you know the answers to all the questions I asked above (and then some), but not before.

Windows vs Linux

[ka9dgx]

There is a heavy percieved bias against Windows, for various reasons here on /., and this story points out the difference between that perception and the reality of the situation.

While many of us don't like Microsoft, for various reasons, we still use it, many of us exclusively. The network you manage is more important than you realize. You're being paid to reduce uncertainty for the users. It's not fair to your users to increase that uncertainty by orders of magnitude unless there is a massive benefit to the users. Replacement of Office with equivalent functionality is NOT a benefit, as it will only frustrate your users everytime anything unexpected happens. The users will also then have you to personally blame when ANYTHING goes wrong, reguardless of its relevance to the OS switch.

If you want to learn some Linux, get an older machine, put two newer NICs in it, and set up an IPcop firewall. You can use a web browser to configure it once it's installed, and peek around inside it via the console. This gives you the benefit of helping to protect the users, while giving you a linux box to manage.

Don't subject your users to your whim. It may suck for you, but Windows is what the users expect, and can cope with, slashdot bias be damned.

Feel free to ignore my advice, but you might want to update your resume if you do so.

--Mike--

Don't... Yet

[zhez]

My best advice is to get yourself a small budget using the TOC arguments that keep flying past here on Slashdot.

What to buy:
1. Two or three used machines (P3s or so). One for a file/print server and two dummy workstations to connect to it.
2. Copies of two or three different distros that are specifically designed for this kind of thing.

I've chosen Xandros Business for my office here (I'm the CTO) since it still lets us run some small windows apps and is very "windows-ish" in feel. Read: Easy to switch. But you should compare at least two side by side. Maybe also Redhat or Suse. Make sure there is official support or else you'll tear out all of your hair while learning the ropes. Community supported distros are only good if you are already familiar with Linux on the desktop and know where to begin looking for help.

Basically, it shouldn't be more than $1-2k for a test. Then you can run internal focus groups, try opening critical office documents in OO etc.

Lastly, some companies will give away one or two copies if you ask them nicely. Xandros gave me 3 copies of Business Ed 2.0 at a trade show because I asked. They were giving Community Ed to everyone else. It convinced my team to switch and we bought the rest of the copies after we were sure.

Im all for Linux

[SlongNY]

Hello,


As the other posted mentioned, dont even think of doing it Live before you know how to do it. Install it at home, setup a PDC/Fileserver first. Setup a desktop to connect to the server, etc. I have used linux for 5 years personally, and didnt attempt to use it for my business until 2 years ago. I had 3 years of exp. with it before i used it on live servers.

Again, as a avid linux user, i always like to suggest it as a option. However in your case, the money is spent, the network works.

Dont fix whats not broken.

SuSE Pro

[MrResistor]

Since nobody else seems interested in posting anything actually helpful, I guess I will.

Buy SuSE Profesional, the full boxed set. For roughly $100, that will get you most of what you need in order to install, run, and learn to use Linux. My personal observation has been that people who start with SuSE stick with Linux. If you really want to be an expert on Linux, there are other distros that you should consider switching to after you're comfortable on SuSE, but I wouldn't start a new user out on anything else.

Before you switch ANY workstations, you'll want to switch the server. Before you do that you need to do 2 things: learn to use *nix (I highly recomend taking at least "Intro to Unix" at your local JC), and verify that you won't be losing any functionality your users care about in the switch (maybe you don't use any of the unique features of Exchange, maybe you do; only you know).

IFF you can switch the server, only then should you even consider switching the workstations. The right way to do this is to start with the apps. Firefox is probably the easiest first step, then maybe email (Thunderbird or Evolution, probably), then OpenOffice (honestly, if everyone inside your company is using OOo, you're better off in the long run having them save stuff in the default format, and just teaching them how to convert stuff if they need to send it to the outside world for any reason), and finally any job specific apps your employees use. ONLY after all apps required for people to do their jobs have been replaced, and proven to be functional, should you even consider switching the workstations to Linux On the bright side, though, at this point no one will care what OS they're running, and some of them might not even notice.

In short, starting from the point you're currently at, expect it to be AT LEAST a year before you're ready to start considering the server swap, and if all goes well AT LEAST another year before the workstations are switched over. YMMV, mostly depending on how much you apply yourself to learning to use and administer Linux. It isn't easy, but it is worth the effort.

I don't quite get it.

[kosmosik]

I don't get it - you have already everything in place, you have server which - I assume - works. You have workstations which work but have some problems with malware (go fix those problems then). You have already paid for everything - so how exactly demolishing your current setup and installing Linux everywhere can work for you?

1. It won't make you save money (on the short run) since you already bought everythink and Linux is not free as in beer also.
2. It will be a lot of learning for you - without proper experience and knowledge your Linux installation will fail miserably - and spending time on learning, gaining experience also costs.
3. Linux is not suitable everywhere (as any other operating system) - it works well in some areas. It does not work in others. For example for common office desktop use it simply lacks applications - sure there is office suite, mail client and web browser - but this is not all that common office needs - go examine what your users need to do their job done and check if there are some Linux equivalents - also mind that Linux software can also be costly (as in money).

So with that in mind I think it is, a completely retarded idea to convert evertyhing to Linux.

What I can suggest you is to pop in one Linux server box to you existing network - integrate it with Windows Networking and see what this baby can do for you. Constantly add features to it. Explore the world of its software - learn what can be done with Linux. Think of any purposes for this system, f.e.:

1. File/print server.
2. Web/application server.
3. Database server.
4. Backup server.
5. Remote access server.
6. Mail server.
7. Lowend net infrastructure server (DNS/DHCP).

Etc. etc. - see if any those roles can help you in your work, can help you save money and so on. Then after some time you will be able to *extend* your network, pop in some other Linux boxes, maybe even on desktop. And make Linux work for you where it is best suitable - but you must know yourself. Linux is completely different world.

And also for more concrete information consult your distribution documentation - and for learning I think the best is task based aproach - so you give yourself a task and then explore to allow you to accomplish it. Like set up a web server.

I would throw in a few more thoughts

[jd]

First, absolutely and most definitely install as many administration GUI tools as you think are necessary - plus an additional 10%. Linux configuration is very powerful, but precisely for that reason it is NOT something the unwary should attempt. Webmin/Usermin or Linuxconf are good places to start.

Second, DON'T try to convert everything in one day. Try to reproduce the working conditions of one computer - preferably a non-critical one - in every detail, just using Linux. Once you've done that, and you've ironed out the problems by letting a user try it, THEN move onto the next one.

Third, do the server the same way - reproduce the system under Linux on a different machine and work on making it stable. THEN try letting a guinea-pig user run connections to it. Once you're satisfied, migrate users over but not until.

To reproduce the server, you'll probably need an Exchange clone. There are several (eg: Open Exchange is a clone down to the protocol, eGroupWare does many of the same functions but isn't a drop-in replacement) and if you shop around you'll find something that'll work for you. But this is exactly why you need burn-in time - you need to KNOW what will work for you, not just think you know.

Re:Troll.

[Master of Transhuman]

I suspect the only troll here is you.

"Throwing out the existing config in order to "save money" is wrongheaded."

Really? Tell your boss that. He'll kill you. Your boss LIVES for saving money - even if he can't and everything he does turns to shit and expense.

Proactive patching for viruses? Dumb Windows idea. Read Marcus Ranum's recent rant mentioned here a day or so ago. It's bad enough you have to patch Linux and its apps for REAL vulnerabilities without having to patch for malware, too.

This guy is correct, if his motivation is a little imprecise. Dumping Windows and switching to all OSS software is the smartest thing he could do, even if the conversion costs for the first year are as much as Windows costs for the last FIVE years. Why? Because otherwise he'll be paying for Windows fuckups for the NEXT TEN YEARS, that's why. And sooner or later that WILL with mathematical certainty exceed the cost of conversion and any ongoing cost of maintenance.