This quick guide is based on true story, and is contributed to the community with the intent to be helpful to the department, office and enterprise Managers, network Administrators and support Staff, and Small Business Owners.
This guide provides general guidelines and ideas for migration path that provides you a non-interrupted work of all your staff during the migration process, moreover it shows some important psychological aspects of such migration so that you both ensure the successful migration from human point of view (your staff) and at the same time extent of their skills so that they are MORE comfortable with their new environment.
Authors’ oppinion about GPL and BSD is his personal vision, but the migration steps are not contradictive with other philosophy (GPL for example) and you may still follow this guidelines although ignoring the BSD-related parts. Not all technical aspects are 100% covered, however such path is proven working (based on true story), so you can advice for further details with a support team, network administrator or rely on your UNIX skills.
1. Choosing the right UNIX.
There are lots of options for choosing UNIX. Most of them coming from the OpenSource world. In most cases a commercial UNIX is more incompatible or hard to build the tons of free apps coming from the OpenSource world. Many of this apps are helpful so you would either have to spend a lot of time of hacking to rebuild your commercial UNIX where you could possibly damage something and lose the commercial support you paid for. OpenSource UNIXes are therefore more friendly to the free apps.
When choosing your UNIX you must consider the ability to update easy your apps and have something to keep the integrity of yuor system (or keep such integrity yourself). If you always download the stuff yourself and compile you are risking the system unless you are willing to spend lot of time on regular basis to keep system integrity yourself.
RPM-like binary is a good option however you are dependent on distro compile options and availability of all packages. RPM source sounds better, but not best, and again the availability problem.
Here you find BSDs and especially FreeBSD with richest ports collection on the net. There are also a couple of Linuxes out there like Gentoo that copied what BSD did. If you have some legacy linux experience already it is a good time to switch to something more real like BSD UNIX. Some people may feel trolling here, so I leave this option to you to choose between the pioneer in ports and the real UNIX – BSDs or the UNIX immitation – Linux.
If you don’t have UNIX experience choose Linux it will be a easy option, choose a distro with RPMs because they will take care of everything.
Author note: People adviced that probably for Linux newbie probably Debian with aptget would be also an easy option, however author does not 100% agree with this because a newbie would not feel comfortable with console while strong RPM based distributions (like RedHat and Suse) have GUI utilities for automatic and manual upgrade with just point-and-click manipulation.
2. Preparing the migration.
Make a local file server – a UNIX based one with Samba. Create accounts for all users, and have them save all files in this location, best is if you apply a rule that no file exists that is saved on local machine except if it’s just something temporary. It is imporant to understand that in long term such centralization of your data provides you with more flexibility and better organization of your files, also such concept will give you better mobility and scalability because one user can always sit on another machine or you can always add a new machine to your office and start using it with little or no configuration.
Prepare a backup script to be safe if something goes wrong with your server, if you cannot do this have one admin to do it for you. You also will need a old machine to serve as your router, here FreeBSD wins. A old box with FreeBSD will also provide you a built-in and proven best firewall and VPN for your network, with easy configuration (at least easier than Linux) scripts for your Firewall, VPN or NAT. Some people just prefer this little devices that serve as router+switch + eventually wireless, however a old machine + BSD is a better option if you want to have better control.
Users that don’t understand anything about console, scripts and UNIX at all can continue operating on their XP boxes, but will be ready at later point to migrate to Mac or other UNIX like FreeBSD (if you are good enough to configure the desktop to be enough familiar and install familiar apps (kde+moz+OOo helps)).
For other users that have some programming experience something better can be done for them. Install cygwin, putty and WinSCP, get them familiar with this apps. Generally if they need to do a console job on their local machine they will start using cygwin (instead of MS Command Prompt), if they need to do a console job on the server – use putty. Then if they need to transfer to directory that is not shared with Samba – use WinSCP, note that once you show them the ‘scp’ command after a certain time they will prefer scp instead of WinSCP.
Other than that, something that applies for all users, start using OpenOffice and Netscape (last version), this apps are already available on UNIX and once they get comfortable with them they are ready. You can also make a research exactly what apps do your users use and check the availability of similar applications for UNIX.
3. Acutal migration
There is nothing special to do once former XP users already are using Netscape, OpenOffice, putty and cygwin. Once they sit on a real UNIX machine they will feel much more comfortable. Moreover konqueror will offer them features that are very UNIX friendly while Explorer didn’t – for example they can set permissions by accessing file properties, or they can do create symbolik links (‘ln -s’) by just dragging the dir in konqueror. Users are much happier with a real UNIX instead of the UNIX simulation ran on former XP box – konsole is better than putty, konqueror is better than explorer, Mozilla is same as Netscape, OpenOffice is exact the same, and even more on UNIX they would have much more apps – there are tons of free apps on UNIX.
4. What’s next
Start thinking how to migrate other users to UNIX. For some of them may be the only option will be Mac since it’s right now the only UNIX with full set of apps that XP users use. Probably in long term strong commercial apps like Photoshop, Corel or Flash currently supported only for one UNIX platform – Mac, would start supporting other UNIXes.
Also you can help your customers to start migration, give them advices to use Netscape (strnong points are that this is business-branded version of Mozilla, has much better security compared to IE and has tabs), also OpenOffice has some stong points (it’s free, it supports more formats than MS, it makes PDF (you have first to explain why PDF is good format for documents)). At later point you can have your customers with small router and a local UNIX Samba server if they have more than 3 machines.
The direction is clear UNIX everything!
5. Optional step – Motivation
If you are a GPL fan you replace everything ‘UNIX’ in this article with ‘Linux’, and you can explain to people why GPL is the best to express such motivation to migrate.
However if you are BSD & UNIX fan you would explain to people about the >30 history of UNIX. You would explain why GPL is not good. You would explain why UNIX is reliabe, proven and so on, you could also mention about BSD, X, Apache and MIT licenses and the major difference to GPL. You could also exlain about the unix heritage of BSD, etc, etc whoever prefers a real UNIX compared to UNIX immitation knows the story.
6. Background
All this is real story. I was planning to add a X server running on the XP box to ease the migration, but I found this irrelevant since KDE is very friendly these days. The only problem we had with KDE was that in version 3.1.4 on FreeBSD 4.9 the folders in konqueror tree are “sticking” to the mouse cursor when you click them, however after recompiling it it was fine.
7. Notes for follow up articles and future work
Expect more articles like this, detailing on certain aspects and points of the migration. Author is commited to advocate UNIX migration and will contribute more material. Three important subjects would be covered in future articles:
1) Detailization of migration processes and tech-related tips for configuration (including helpful links of howtos and manuals)
2) Advocating and motivation articles on why to choose UNIX, why to migrate, and how to convince your staff and customers
3) BSD&UNIX phylosophy and topics on Free Market + Open Srouce
About the Author:
Anton Velev is long term BSD and UNIX advocate and old KDE user. Although he didn’t contribute any code to OSS project, he spreads UNIX revolution by posting articles and forum posts, migrating and training his staff and customers and advicing other people to migrate and how to do so. He is commited to continue contribution of his advocating articles and for future to fund and contribute open source projects with code, documentation and guides/tips/tutorials/howtos.
If you would like to see your thoughts or experiences with technology published, please consider writing an article for OSNews.
what should I migrate my exchange server to?
I’m sorry to say that, but this is the worst article I’ve ever read on OSnews. It is full of typos and uses a *very* bad english.
Besides, his article is not detailed enough to cover such a complex topic. I feel I have not had any benefit from reading it. But that’s another story and others might just get what they’ve been expecting.
Eugenia, you’re doing a great job and I’m really appreciating it, but please, be more careful in the selection of articles. It will benefit the reputation of OSnews.
You are very welcome to sit down and proof read the article and send it back to us to update it.
You should not expect nothing more from volunteers on this site. I already do a lot of work here.
I use both FreeBSD and Linux. Comments implying that the users were already familiar with cross-platform apps such as OO were interesting; I think this article would have been better if it stated the preparation for the migration a bit more explicitely.
All the arguing about how Linux is an “imitation” and not real, and complaining that it’s easier, and GPL-bashing detract from the good points the article makes. I entirely fail to see why an “enterprise migration” needs to convince all the users that the GPL is evil.
“However if you are BSD & UNIX fan you would explain to people about the >30 history of UNIX.”
… which did not start as BSD. It started as a proprietary system at at&t; it just was not treated as strongly proprietary until the Bells broke up and could enter the computer business.
“You would explain why GPL is not good.”
This has _nothing_ to do with a unix migration. Furthermore, the author appears to advocate KDE, which is….. GPL.
“You would explain why UNIX is reliabe, proven and so on,”
Fine.
“you could also mention about BSD, X, Apache and MIT licenses and the major difference to GPL.”
Unless they’re programming, their eyes will probably glaze over.
“You could also exlain about the unix heritage of BSD, etc, etc whoever prefers a real UNIX compared to UNIX immitation knows the story.”
Sure, why not talk about the lawsuit too?
BSD and Linux both have their strengths. BSD has a more liberal license, and is older. That doesn’t make it better.
The author’s constant anti-GPL comments, combined with supporting KDE, leave me perplexed. I’d think he’d at least support gnome or some other WM which wasn’t GPL’d, given how virulent he is about it.
If you want articles proofread, I’ll do it. It might take me a day or two, depending on how busy I am.
Yeah, but you need to email our team, not reply here. This is already off topic, and you don’t even include an email address. So, whoever wants to help out, just email our team.
The article content is fine and its not a bad read, but he should have just said FreeBSD instead of Unix.
the article should be called.. how to migrate 1 or 2 people in an office of 3 to my favorite leet hax0r platform.
I didn’t get the point with motivation – GPL, BSD – does it means that I should migrate from gpl-*nix to bsd-*nix or vice versa? Or maybe author imply that one would migrate from Windows? Then write about motivation to do that!
I have to make some notes to clear misunderstandings and reply people who misunderstood some of the points:
1) “what should I migrate my exchange server to?”
First, there is suseexchange, a very strong commercial replacement, then also there is another exchange compatible suite for MacOS X Server if you want to migrate right away. There is also Lotus Domino from IBM that also has migration scripts and wizards to migrate from almost everything. This applies if you want to rely on commercial support, however if you are willing to do so you can also combine yourself many different suites available for UNIX (sendmail, qmail, openldap, etc). Important point is if you first migrate your client software so that it’s not dependent on exchange.
2)”..uses a *very* bad english..”
I didn’t check with the spellchecker, I am sorry. I hope you can undersntand it.
3) “.. but he should have just said FreeBSD instead of Unix.”
I am not suggesting that user must choose exactly FreeBSD. Please read carefully, you can find/replce it with your favorite distribution. And you know BSD is UNIX, while GNU/Linux is not.
4) “More technical details and less politics would be nice.”
I agree. I will continue with more technical details in next article. I am BSD advocate, although I suggest that GPL software is a good start for newbies, it’s easeier for learning, but once you learn from Linux you would be ready to switch to UNIX (being it BSD, Solaris, AIX, Mac or other UNIX).
5) “the article should be called.. how to migrate 1 or 2 people in an office of 3 to my favorite leet hax0r platform”
Totally disagree here. Such migration path is extremely scalable since your data is initially centralized. You also cannot disagree with strong relyability of a FreeBSD system as a server platform. What’s good news is that it’s perfect for desktop too. Back to old days when I was trying to find a way to delete Linux and use BSD only the two apps that I was unable to get on BSD were OpenOffice and Java, now they work fine.
6) “Motivation?”
Yes, definitely migration from XP to UNIX (or UNIX immitation). I gave options here, you choose what would you migrate to. However it is important to express a certain level of motivation based on the freedom that UNIX provided (being it the “protected” freedom with GPL, or the total freedom with BSD).
While I use and enjoy an RPM-based distro, I find it hard to swallow the article’s recommendation to “choose a distro with RPMs because they will take care of everything.”
RPMs (when dealing with them directly, without yum or urpmi or yast) are notoriously difficult to work out dependencies. This point should have been made clear in the article, as well as better descriptions of the other methods (though BSD port was covered well) such as apt, so the reader could enter the unix world with open eyes (or at least without an unfounded bias toward RPMs).
Mark
(2) Anton, thanks for replying to my comment. I appreciate this move since not everybody takes the time or is feeling unnecessarily offended.
By the way, I am not a native english speaker either, so I can perfectly understand the difficulties.
Mixed feelings about this article.
Being a BSD advocate is fine, but it doesn’t excuse you from the Linux bashing you do in this article. It’s an interresting topic you’re trying to discuss, but it seems you’re taking it too personal.
UNIX imitation? Gimme a break. UNIX derivative, or even UNIX like would be much better digestible than your drivel about how BSD is a “real” UNIX. Define what a “real” UNIX is first, and then explain how Linux doesn’t but BSD does fit that definition, and then you *might* have a point. Without doing so, you’re just a troll, go bug slashdot instead.
Also, you’re rather light on the details of what your starting position is, or even what the goal of this whole excercise is. You seem to rake some random thoughts together, and list them incoherently as a context to some BSD advocacy.
If I were to be editor for this article, I’d ask you to rewrite it. Keep the amount of flamebait to tolerable levels, run a spellchecker on it, and also to make it a conclusive article. Ending it with a vague promise for more details, especially after the trolling and flamebaiting you do, isn’t going to draw more readers to your story.
Excuse my bluntness, but this whole thing just screams “amateur”, and not only for the writing articles part.
paste from the content:
“If you don’t have UNIX experience choose Linux it will be a easy option, choose a distro with RPMs because they will take care of everything.
Author note: People adviced that probably for Linux newbie probably Debian with aptget would be also an easy option, however author does not 100% agree with this because a newbie would not feel comfortable with console while strong RPM based distributions (like RedHat and Suse) have GUI utilities for automatic and manual upgrade with just point-and-click manipulation.”
Yes, I am exactly talking about GUI applications like Yast that will help a new Linux user to update his system.
The author mentions setting up a fileserver, and encouraging users to save files centrally on the server instead of on their local machines. That is very good advice.
However, with Samba 3 you can do a lot better than that! You can integrate a Samba 3 server in a 2K or 2K3 Active Directory environment, even have it serve as an AD domain controller. That includes implementing roaming profiles, so that users’ “My Documents” directory, their applications settings, their IE booksmarks, their desktop wallpaper etc. are all stored on the Samba server. Users can log into the domain from any workstation and all their files and settings are right there.
You can also use Windows NTFS ACL-style permissions on the Samba server. That’ll give you much more precise security control over files than plain vanilla UNIX u-g-a permissions.
As for Exchange compatibility … Suse Openexchange is nice *if* you need to use Outlook as your e-mail/calendar/directory client. If you’re not tied to Outlook, you’re better off using GroupWise (at least on the Novell side). If you are thinking about “rolling your own” Exchange replacement with qmail, OpenLDAP etc., well, good luck! E-mail me your config when you get it working!
Many of this apps are helpful so you would either have to spend a lot of time of hacking to rebuild your commercial UNIX where you could possibly damage something and lose the commercial support you paid for.
This is the worst reason to migrate an enterprise to OSS. Not only is it completely opposite of what 99% of businesses out there use commercial UNIX for, believe it or not most vendors are more that happy to accommodate you with whatever you are attempting to use their platform for. This idea that you have to alter the source of the operating system in order to faclitate some obsure need is ludicrous. Secondly converting entirely to one platform or another is not good business sense. A little thing called vendor lock in can come back to bite you in the arse.
“Mixed feelings about this article.
Being a BSD advocate is fine, but it doesn’t excuse you from the Linux bashing you do in this article. It’s an interresting topic you’re trying to discuss, but it seems you’re taking it too personal.
UNIX imitation? Gim…”
Although I am BSD advocate, I must add that I have been extremely tolerant to GPL/GNU/Linux flavors, stating not just once that it is your oppinion based on your UNIX skills or “political” affiliation. Talking about GNU/Linux being UNIX or not you know what GNU means (GNU is Not UNIX), more over we know the way linux was written from the scratch by volunteer hackers on the net.
While BSD was written in academic/professional environment with a solid organization and building directly over the original UNIX, while also improving it. Please note that BSD code exists today in every commercial UNIX, even in NT kernel.
check this:
http://www.freebsd.org/cgi/cvsweb.cgi/~checkout~/src/share/misc/bsd…
and this:
http://www.opensource.org/timeline.png
Learn something about the history of the Operating Systems, check who invented the first internet standarts…
For sure Linux is more and more becoming UNIX, but check where Linux took the code from in order to move closer to UNIX.
And remember that Linux system is Linux kernel + mishmash of GNU apps. GNU (which stands for GNU is Not UNIX) is definitely a UNIX decedent, the Linux kernel is just build from the scratch (else SCO would win the lawsuit). BSDs however provide entire System – a kernel + unix tools all guaranteed working together as a whole, made or integrated by their teams.
This is getting too off the topic but anyway think about it.
I Agree. there are always going to be some app that is dependent on win32……other than that you should be at least considering Linux..unless your scared or Lazy.
-nX
I Agree. there are always going to be some app that is dependent on win32……other than that you should be at least considering Linux..unless your scared or Lazy.
-nX
I have considered it and use it at home for server duties i don’t use it for Desktop systems since i have a HP B2000 workstation and a SUN Ultra 10 workstation to choose from for Unix related tasks. I refuse to recommend Linux as a solution to any business owners while commercial Unix is still available. I don’t want to get into the Unix vs Linux discussion, this is a personal choice of mine and fellow System Administrators who work in large business IT environments.
“The author mentions setting up a fileserver, and encouraging users to save files centrally on the server instead of on their local machines. That is very good advice… …Active Directory environment, even have it serve as an AD domain controller. That includes implementing roaming profiles, so that users’ “My Documents” directory, their applications settings, their IE booksmarks, their desktop wallpaper etc… …Windows NTFS ACL-style permissions on the Samba server.”
Thanks, JH, Yes, with Samba3 and ACL you can do a lot more when dealing with samba server and having your users saving to it, but there is one important thing that you cannot do from any XP box never mind what Samba you use. It’s making the symbolic links, which turns out to be one of the easiest ways to organize files in UNIX environment. Windows user with Explorer just sees folders (and not links) but is happy to have the documents “auto-organized”, however is impossible to “construct” the “auto-organization” himself.
Talking about Samba and windows boxes, you can make the client windows boxes “diskless” (i.e. docs on server, apps on server too, os to boot from network too), may be only a small hard drive for swap or just a big memory.
What’s good in this strategy is that work is not interrupted at all, and the migration is smooth.
“Although I am BSD advocate, I must add that I have been extremely tolerant to GPL/GNU/Linux flavors…
blah, blah, justification, blah, blah
…a whole, made or integrated by their teams.”
I know perfectly well where Linux and BSD descend from, thank you.
You’re the one making a point of it, trying to justify why BSD is “UNIX” and Linux is not. Does it matter? No, not one iota. What you intend to achieve with the whole transition from whatever to “UNIX” can be done with any free “UNIX”, and maybe even with a non-free “UNIX” if you want to bother with their “quirks” in running the tools you need on them. That’s not my beef with the article, my beef is that you deem BSD more worthy of functioning as a glorified fileserver for Windows clients because it’s more “UNIX”. I call major bullshit on that, because it is, indeed, bullshit.
To be honest, I wonder what you actually try to achieve with this article. Is it to make more people think about switching from running fileservers on Win* to running fileservers on *nix? Well, with the tone you’re setting in this piece you’re not likely to swing over your target audience. It’s woven through with your subjective personal preference and opinions, and rather low on justification and detail.
Sorry, but BSD advocates bitching and moaning at their perceived “competitor” Linux isn’t going to impress anyone, especially not the people making the decisions.
Author note: People adviced that probably for Linux newbie probably Debian with aptget would be also an easy option, however author does not 100% agree with this because a newbie would not feel comfortable with console while strong RPM based distributions (like RedHat and Suse) have GUI utilities for automatic and manual upgrade with just point-and-click manipulation.”
Yes, I am exactly talking about GUI applications like Yast that will help a new Linux user to update his system.
In the case of apt, Synaptic is a very easy-to-use GUI frontend to apt.
Sorry, but BSD advocates bitching and moaning at their perceived “competitor” Linux isn’t going to impress anyone, especially not the people making the decisions.
Very well said!
Okay, all you admins out there: have you not found that the “best” solution is the one that gives your end users whatever management says has to be delivered? We all know what that means. Sooner or later, you’re going to be running a mixed OS network, and if you’re smart, you’ll use the best tool for the job. Many of us have found that any webserver on the DMZ other than Apache is – – well – – how would one explain any other choice? Seriously, though, I’ve been a Linux freak since 1995, but I have no delusions whatsoever about converting my mostly Windows network in its entirety to Linux, Unix, Mac, or anything else. If you’re working in a specialized industry (and who isn’t), then you know that a lot of apps continue to be available only on the MS Windows platform, and you’re lucky if the database backend can be anything other than Oracle or MS SQL. I don’t even dream of an all-*nix network because it ain’t gonna happen. But, here and there, I sneak in a non-Windows server. And what’s this about trying to make Samba work as well as a Windows Print/File server? I’ve been there, and I soon concluded that for all it does, a Windows 2000/2003 license isn’t THAT expensive. Unless you’re having to buy CALs for everyone, of course . . . .
Assuming I will recieve a barrage of negatvive responses based on my ip…
I won’t repeat what others have said above regarding lexical correctness. Suffice to say, I belive professionalism is important in public communications; and a part of developing professional content is accurate use of the dissemination language.
RE: Application Compatibility
I was dissappointed by the single paragraph description of the actual migration of users off of Windows, onto UNIX. I expected a more well thought-through migration plan, including consideration of machine cost, hardware compatibility, deployment/downtime cost (how long will my users be offline? Time is money, after all), and probably most importantly what I can do to provide application support.
Of course by application support I do not mean email and web access. What I mean is, how are they going to continue using the cutsom schedluing applaiction for physical therapy patients which is Access/JET based? How are they going to continue to run their complex arithmetic analyses which are dependent on a thrid party COM based Add-In specifically designed for accountants in the greater Fresno area? This software is nothing to sneer at. These small businesses have real investments, both time and money in these tools. For them, the expense of repurchasing, retraining (assuming an equivalent tool is available), and redeploying the software are the primary hurdles they face when considering a migration.
In some cases, these small businesses are exactly one week ahead of the game. Losing a week of productivity to this operation could well cost them their livelihood.
Some of you are probably shaking your head in denial of the scenarios I’ve presented. You are of course free to believe me or not, as you choose; though in fact I have a great deal of experience dealing with clients of this sort, and that they are very common in the American businessplace (I don’t know much about European small business).
Show me the $$$…
The above not withstanding, I do not suggest that small businesses cannot be migrated to UNIX/Linux based platforms. In fact, I would be very interested in reading an article written by someone who has been doing this very thing, and has found reliable methods for dealig with the issue of custom solutions.
I can say with a certainty that I could not do my job without Windows. I am far too dependent on Windows-only custom tools to be able to do such a thing. While I’m sure you are not suprised to hear that this is my case, I submit that it is also the case for employess at most business people, from the Fortune 500 to the 911 call center in your region. I’ve talked with a lot of customers that are in this situation.
In the end, it all comes down to money. Small and large business owners alike want to do whatever is the cheapest thing that gets their job done. The real way to present an attractive migration plan is to prove that it will be worth their while monitarily to change they way they work now.
My $.02
-Zac.
P.S. I also don’t agree with the assertion that users will find conqueror/KDE easier to use than what they’re used to. Business users do not want to explore the software (I have labratory gathered staticstical proof of this); they just want to learn something once and then do it exactly the same way until the end of time.
I totally agree to this.
Since my work is focused on usability and user experience, it is absolutely true that for most people in business and private (apart from computer geeks), it is a pain in the anus to learn something on the computer. They want to get their job done, period.
The success of a software is largely dependent on the usability. If a software is poorly written, but has a great and intuitive user interface, people tend to prefer this bad software instead of a great written software with poor usability.
Thus, forcing users to re-learn something just to get the same job done that they could get done before with the old thing, is a clear no-go (and will most likely fail to succeed).
Sit someone, who has never used a computer before, in front of GNOME, and they will hardly get used to windows.
If someones computing experience started with windows, they are going to have difficulties getting used to something different.
My sentiments exactly.
Why spend all this time and effort replacing a simple Windows fileserver which does the job with, well, a headache?
What does a Windows server license cost, depends, but I’m sure it can’t be more than a $1000/day consultant hacking away at some hackjob solution for a few weeks. And why do you need a Windows fileserver? Because there’s apparently already Windows clients connecting to it, so that immediately takes the CAL argument out of the equation (I do assume a reasonably sized shop here, with proper MS licensing). So where’s the advantage? Support is nil (unless you manage to keep that same hack consultant on call 24/7). Vendors won’t burn their paws on it (especially when you mention “Oh yeah, and it runs BSD”, with linux you might have had a chance). And your “cheap” solution becomes very expensive, very fast.
You know what the problem is with all these techies writing articles about how the whole world should convert to OSS solutions tomorrow? They have no friggen clue how a reasobaly large IT organisation is run and how the decision makers (your manager’s manager) come to their decisions.
Does a decision maker really care which OS drives their fileservers? Naw. They care about vendor support, resiliency, recoverability, availability, contigency, maintainability, administrability, *bility. Hardware and software cost is just a tiny part of any consideration when looking at the big picture.
Is there any reason to replace a Windows based fileserver with a Samba fileserver? Maybe, but does it make sense when looking at the larger picture? Hardly. And would a BSD solution have any chance of seeing the light of day in any sensible IT organisation. No way. And that’s not to be mean, but that’s a sense of reality. Where are the certified professional for BSD? Where is the vendor backing? Where is the garuanteed working hardware from industry partners (and don’t even try to mention BSD Hardware Inc. from Backwater, USA. I mean HP, IBM, Dell, the usual suspects…and what about unimportant players like, oh, Getronics to actually support the whole infrastructure, including emergency recovery, etc). In other words, who do I turn to when the sh*t hits the fan and I have to get things working again by 5pm *today*?
Write about that sort of issues, and you *might* get your target audience interested. What Anton has now is nothing more than a written pipe dream for zealots and advocates who think because they manage to “convert” a few users to a Samba server have solved the world’s IT problems.
And that’s only about the sanity of replacing Windows file and print serving with a BSD/Samba solution. I hardly even scratched the surface.
“… I wonder what you actually try to achieve with this article …”
This is would be a much longer topic if I try to explain you right now.. You are trying to switch to offtopic subject, while I mentioned several times that BSD opinion is my advocated opinion and people that prefer other UNIX(like) OS may just ignore it.
Please ignore whatever you don’t like (you can even just find and replace in your text editor everything that is UNIX or FreeBSD with Linux and GNU) and reread the article again. Read carefully please, you will find out that the path I am suggesting is a path that provides non-interrupted work of your users while progressively migrate. Whoever needs to migrate may find something useful, moreover if I provide more tech related material in next articles.
Nobody is forsing you to migrate, best is if you are just a multiplatform and feel comfortable with any platform. However sharing such experience and contributing it to whoever finds it helpful I find as a good thing. Why don’t you contribute your experience?
About UNIX cert stuff: Yes, FreeBSD does not have UNIX cert but I don’t care, and I don’t care if MacOS, AIX, Sun, SCO have or not too. What I do care is that they are all UNIXes and they share this heritage and have grown from same code base and built in same professional and academic environment coming since the ages of AT&T, Berkeley, MIT, NCSA, University of Illinois and etc.
“Assuming I will recieve a barrage of negatvive responses based on my ip…”
You make good points. I like you already 🙂
If people dismiss your comments solely based on who you work for (assuming you do work for “the enemy” 😉 you can safely ignore them. No use trying to talk to people who only listen to what they want to hear anyway.
Welcome, and I look forward to your insights.
Not that I actually matter in any way, mind, I’m just trying to make you feel comfortable before I go for the throat…*cough* 😉
Ok, many people ask why would samba be better than Windows fileserver, let me answer from practical point of view what FreeBSD+Samba gives more:
1) Samba is proven to outperform the original Windows SMB
2) Samba has better configurability
3) the UNIX file system supports natively the Symbolic links which gives you in return better organization of the files, moreover done automatically
Overall about UNIX, it’s more secure, you have better control over the system, you can read the code, you have freedom choose between a variety of vendors (distros,projects,patch-commiters) standarts-compilant product (in fact it’s a free market driven open competition in the UNIX & Open Source world).
Other than that if we exclude the real money costs calculation there still remains the always underestimated cost of freedom to choose and customize your product (or even extend it yourself).
Although it might not be complete, I noticed OpenChange (BSD licensed) just recently: http://www.openchange.org/. It is attempting to be a replacement for Exchange.
You know what the problem is with all these techies writing articles about how the whole world should convert to OSS solutions tomorrow? They have no friggen clue how a reasobaly large IT organisation is run and how the decision makers (your manager’s manager) come to their decisions.
Does a decision maker really care which OS drives their fileservers? Naw. They care about vendor support, resiliency, recoverability, availability, contigency, maintainability, administrability, *bility. Hardware and software cost is just a tiny part of any consideration when looking at the big picture.
<sniff> brings a tear to my eye to finally have someone else that understands how IT in business operates on this website. Thanks for saving me the time of typing the same facts over again.
Why spend all this time and effort replacing a simple Windows fileserver which does the job with, well, a headache?
What does a Windows server license cost, depends, but I’m sure it can’t be more than a $1000/day consultant hacking away at some hackjob solution for a few weeks.
A few weeks? What are you talking about? As someone who works for an actual professional system integrator … I can set up a secure Samba file server in less time than it takes to just install 2K3 and apply all the updates from Microsoft (which my company does on a regular basis as well … we’re no OSS zealots, we’re just trying to get the job done). I’m talking maybe 45 minutes for the OS install and copying over the config, maybe 15 minutes to register the server on the domain via Kerberos, copy over the smb.conf and test everything out.
That complicated configuration for a AD-authenticating, Winbind-enabled (all files stored using Windows user and group permissions) file server … guess what? It’s not complicated at all. Your smb.conf file apart from any share definitions is like 25 lines total. And with *nix it’s easy for us to mirror configurations across machines.
Why *NIX over Windows for something like a file server? Duh … it’s cheaper, performs better on the same hardware, is faster to setup and configure, scales better, is more scriptable for backups, is secure out of the box, requires vastly less patching and net monitoring, has better uptime …
This isn’t anything new. *NIX has been the large-enterprise backoffice platform of choice for the last 25 years … and now with x86 Linux/BSD/Solaris it’s no longer out of the price range of our SMB customers, which is great. HP, Dell, IBM … every server they sell can be purchased with RHEL or Suse preinstalled.
Again: Such migration path provides you uninterrupted work while migration advances, from and to whatever you want to migrate it will work, you choose the the start and end point.
You pay prefer other UNIXes while this story just worked and is real (the opposite of unrealistic).
<quote>
Ok, many people ask why would samba be better than Windows fileserver, let me answer from practical point of view what FreeBSD+Samba gives more:
1) Samba is proven to outperform the original Windows SMB
</quote>
Raw throughput is a very deceiving factor. If I told you I have 800 users accessing a single fileserver running Windows 2000, and not having any performance complaints, would you believe me? You better, it’s the truth. Why? Because raw, streaming throughput doesn’t matter in a regular office environment where lots of small files live which are accessed very asynchonously.
<quote>
2) Samba has better configurability
</quote>
You will have to explain this one to me. Looking at the entire configuration syntax Samba supports, I see little which Windows 2k+ doesn’t already do natively. Configured or not.
<quote>
3) the UNIX file system supports natively the Symbolic links which gives you in return better organization of the files, moreover done automatically
</quote>
I fail to see relevance. What is “better organisation of files”. There’s a share, there’s files in it, people access those files, end of story. What’s there to organise? And why would symlinks be a crucial factor in that? But maybe I’m missing something important here, please elaborate.
<quote>
Overall about UNIX, it’s more secure, you have better control over the system, you can read the code, you have freedom choose between a variety of vendors (distros,projects,patch-commiters) standarts-compilant product (in fact it’s a free market driven open competition in the UNIX & Open Source world).
</quote>
Security is the only point you could make when choosing UNIX as a solution in this context. But then again, security isn’t just a technical solution, and to only push UNIX forward as “if you run this, you’ll be secure” is sticking closing the eyes for deeper problems which aren’t easily solved by just replacing an OS.
Who is going to look at the OS code, administrators? Why would they want to? That argument only flies in organisations which are run by techies, or which have developers with too much time on their hands. I’d really like to see some numbers on where an argument like this was actually a design goal, and a followup on whether or not anyone actually cares about this afterwards.
Regarding standards, everyone in the real world standardizes to be Microsoft compatible. Even the OSS world. Proof in point, wasn’t it Samba we were discussing? And is it really such a good idea to have a lot of contributors to software? You see how that turns out in the linux camp. An unintegratable mess of applications without a clear development path. And don’t point to me that BSD is better; the same applications which are used on linux are used on BSD too. What’s effectively the difference? The kernel, the way software installs and some other minor details which matter little after things work. All the rest is just as much an unintegratable mess as over on linux. I fear the day someone will tell me that I have to replace XP on 800 desktops with an OSS solution, keeping the same (lack of) functionality as they have now.
<quote>
Other than that if we exclude the real money costs calculation there still remains the always underestimated cost of freedom to choose and customize your product (or even extend it yourself).
</quote>
Again, why would anyone want to? I know little organisations which are willing to hire a developer to modify something which should be a commodity, an OS. That’s admitting the choice to go with said OS was a mistake. If it’s applications you’re talking about, you’re aware that most developers are actually developing against standard software packages, right? COOL:Gen, Oracle, various kinds of accountancy packages, and a whole slew more. Noone *cares* about “productivity” software enough to want to modify it. If it runs acceptably after installation, noone will pull resources together to get someone to hack around in something which should be a commodity.
Freedom is a “nice to have” feature, not a critical design goal.
IMO, of course.
Not sure where this supposed “business” bias against Linux is coming from … we sell Linux solutions to customers all the time.
Yes, if you are an IT administrator proposing a Linux/BSD solution for your company, or a consultant proposing one for a client, of course you have to make the business case. If you are proposing an unsupported OS like BSD or Debian, your company has to have the ability to provide 24/7 uptime. But that’s not a requirement … buy Suse or Redhat and you’ll get much more responsive support from those companies than you’ll get from Microsoft for a couple of 2K3 licenses.
We support Windows, Mac and mixed Windows/Linux environments. The TCO case for Linux is very clear: lower hardware costs, lower software costs, lower manpower costs. The manpower costs boil down like this: even when the initial setup time for Linux is greater, the day-to-day knob-twiddling of keeping a Windows network running (you know, the net monitoring, the patching, the reboots, the emergency calls when all of this happens at night) is what makes the difference.
Just think about the filesystem advantages of *NIX alone … compare a high-performance journaling filesystem like XFS or ReiserFS to NTFS.
Now would I run a Outlook/Exchange shop without a 2K or 2K3 server as my Exchange machine and my main domain controller? No. But file services and stuff like secondary DNS are another story.
And if you don’t need Outlook/Exchange, you don’t need to deal with Windows on the server at all. Which makes life so much easier!
As far as this symlink nonsense goes … you can’t actually create symlinks via Samba. Of course you can SSH into the box and create a symlink, and then on Windows it’ll show up as a complete duplicate file tree … but what IT manager is going to let their regular users SSH into a running file server? And who are these users exactly who even know what SSH is? These are people who don’t even bother with Windows shortcuts, so they’re definitely not going to care what a symlink is.
As I recall, one of the downsides of the Samba solution was that the Samba server had to maintain its own separate list of AD users locally, occasionally syncing with a domain controller through the use of a script. I believe that was part of Winbind and PAM. In my organization, users come and go on a weekly basis, if not daily. I don’t want to keep multiple images of my user database on member servers, and I don’t want to be concerned about whether a file server is out of sync with a DC. Has this changed in Samba/Winbind? Can it work seamlessly in a single sign-on AD environment? And are file permissions still being translated between Unix and NTFS? If a user changes his/her password on the domain, do they still have to connect via some sort of shell session to change it on the Samba server as well? It’s been a while since I played with Samba for anything more than personal network shares on workstations – – I’m curious as to whether the “enterprise” experience has improved.
My first Unix was an HP-UX at work, the second was a SunOS running on Sparc hardware, the third (and my first “personal” unix) was a FreeBSD and now I use exclusively Linux because it is better for desktop usage, your development rate is greater than *BSDs and because GPL is the only way to protect free software from monopolistic companies like Microsoft and others like Apple (who sell proprietary software AND hardware and it doesn’t support free software. They only used the best parts from them and make another system full of proprietary technologies).
Linux is not for newbies only but for everyone who cares about the philosophy of free software.
I’d love to hear about migrations back to Client/server model (Meaning MS Terminal server for MS Office use and a UNIX/Linux server for vertical JAVA/UNIX applications).
Thin clients are getting cheaper nowadays and IMO are the future of computing for business.
Fred says:
And would a BSD solution have any chance of seeing the light of day in any sensible IT organisation. No way.
Maybe this report could be enlightening.
http://news.netcraft.com/archives/2004/06/07/nearly_25_million_acti…
Wow, a 2 page guide to migrating your enterprize to UNIX, and the section on “doing the actual migration” is 3/4 a paragraph long… must be gold
This article was begging for a flaming – you could tell after the first few lines.
Go ezy everyone.
I am sure he had the best intentions and learned some valuable “how to rub Linux and GPL the wrong way” tips. I bet there was no innocent newbie that read it and said to himself “Wow! I didn’t realise that BSD was so much better than Linux!” – I really don’t think so. I think he got the picture and hopefully won’t tell us to get stuffed and never write an article again.
Neither licence — nor any other — are always ideal. To say so or to bash one like this author has is not even good advocacy.
For the record, I like and appreciate many licences and only get annoyed at a few — though it’s not my code to licence!
For example, the GPL is great for keeping any one group from dominating a project. Projects that require cooperation are good candidates for the GPL. It makes normally hostile companies, for example, feel confident that they will not loose out to each other since everyone has to play by the same rules.
BSD or better yet public domain licences are fantastic for anything that deals with a specificiation since the licenced reference design (for example a protocol stack or a data format) can be used by a variety of groups and compatability is given from the very begining. It reduces needless errors in custom implementations that become defacto ‘standards’.
Commercial licences — including commercially licenced BSD or GPLed software — allow some control over how the software is sold, supported, and managed. For example, Red Hat uses a commercial licence so that they are not stuck supporting 100 computers for every box of Enterprise sold.
If you create the code, you can licence it as you wish. If you don’t, your opinion doesn’t matter much — it’s not yours! Licence sniping is just in poor taste and shows a lack of understanding about the licences and the roles they serve.
what should I migrate my exchange server to?
Groupwise, Koolab, Notes, SUN Microsystems SUN One Software Stack, Oracle etc. etc.
Exchange is NOT the be-all and end all of messaging servers; anyone who has ever used Exchaged then moved to something else will tell you that it isn’t until you’ve tried something else when you realise how crappy and overpriced exchange really is.
As I recall, one of the downsides of the Samba solution was that the Samba server had to maintain its own separate list of AD users locally, occasionally syncing with a domain controller through the use of a script. I believe that was part of Winbind and PAM. In my organization, users come and go on a weekly basis, if not daily. I don’t want to keep multiple images of my user database on member servers, and I don’t want to be concerned about whether a file server is out of sync with a DC. Has this changed in Samba/Winbind? Can it work seamlessly in a single sign-on AD environment?
Yes … Winbind works seamlessly in an AD environment under Samba3. Basically, Windows users and groups are created automatically as UNIX users and groups. To see it in action, you can add a user or group on a DC and then do “getent passwd” or “getent group” on the Samba server, and the users and groups show up immediately. They can browse to file or printer shares on the Samba server right away … no login boxes pop up or anything like that. Normally we set up multiuser Samba fileshares based on Windows user groups, with “force group” set to whatever the Windows user group is and “create mode” forced to “ug+rw”.
Also … once PAM is working you can use it for lots of things besides Samba. Like you can edit /etc/pam.d/ssh and allow domain users to SSH into the box using their domain usernames and passwords. And likewise for FTP or X. So it’s single sign-on for more than just file/print sharing if you want.
And Samba/Winbind can take care of creating users’ home directories if they don’t exist … pretty key for roaming profiles if you’re adding users to the domain frequently.
And are file permissions still being translated between Unix and NTFS?
Whatever UNIX permissions are on the files (using the domain users and groups) are applied and take precedence. In addition, via Windows you can right-click on files and change their permissions individually just as you would on a Windows server, provided the “nt acl support” flag is set in smb.conf. However, the POSIX ACLs that Samba uses aren’t exactly the same as NT ACLs, so while you can modify the “rwx” of a file, you can’t set the “Delete Access” bit. And you can’t add users or groups to the ACL for a file or directory if they aren’t already granted some privilege in the [share] definition, you can only take privileges away.
In practice this isn’t something our clients really care about or even want enabled at all … usually they want to force group rw on multiuser shares so that all users with access can change everything.
If a user changes his/her password on the domain, do they still have to connect via some sort of shell session to change it on the Samba server as well? It’s been a while since I played with Samba for anything more than personal network shares on workstations – – I’m curious as to whether the “enterprise” experience has improved.
That’s pretty old-school … if you have Samba set to AD or NT4 Domain authentication you don’t have to worry at all about that. AD/Domain auth means Samba connects to the domain or NT4 PDC on-the-fly to authenticate users when they try to access a file or printer share. With AD auth this works over Kerberos (encrypted), with NT4 it’s unencrypted (and NT4 auth still works on a 2K domain, and even on a 2K3 domain if you turn off packet signing on the ADC).
(You can of course still do machine-level auth with Samba … that’s where you connect by shell and use the smbpasswd command to set your password. It’s the same idea as peer-to-peer file sharing between two Windows machines — you can’t change your password on the remote machine without logging into it. Not something you’d use in an enterprise environment. But since it’s the simplest it’s the install default.)
“You are very welcome to sit down and proof read the article and send it back to us to update it. You should not expect nothing more from volunteers on this site. I already do a lot of work here.”
Its a good idea to do this before the article is published to the public. I propose to create a group which does precisely this before it got pusblished, to do some spell checking, basic fact checking. Create a diff (‘wiki-style’ editting), discuss with author (develop consensus), then post (to the public readership). Though this group should not force any of their ‘vision’ in the author throat or authors article, therefore i suggest: 1) Author’s opinion on content is final. Author is allowed to use the changes in future work for whatever license he wants to. OSNews’ opinion on posting yes/no is final (latter is same as right now). 2) The group focusses on spell-checking, basic fact/error checking and unargumented though loaded points. Group develops this, mails/sends to author the points made by Group. After that, either discussion / trying to reach consensus, or author implements changes. OSNews editors then decide wether the article is posted or not. Group can be (semi-)independant from OSNews or OSNews-editors. People who do this could be drawn from the readership pool.
Rpm can not be compared with apt, only with dpkg. Dpkg is the actual package management tool and apt takes care of dependencies and fetches required files, then dpkg installs them.
So comparing apt and rpm is silly. And if I may note, apt has been ported to rpm based distros too, so if you learn to know how to use apt in rpm based distro, then you’re nearly ready to migrate to debian or ubuntu, since this article is all about migration.
meu isto que eh uma discussao bem feita
>I’m sorry to say that, but this is the worst article I’ve ever read on OSnews. It is full of typos and uses a *very* bad english.
That’s the reason the article is bad? Typos?Get a life.And do that soon
>Besides, his article is not detailed enough to cover such a complex topic. I feel I have not had any benefit from reading it. But that’s another story and others might just get what they’ve been expecting.
Then why did you read it?Nobody’s forcing you.You had no benefit? Who cares,there are some people who MAY benefit.
>Eugenia, you’re doing a great job and I’m really appreciating it, but please, be more careful in the selection of articles. It will benefit the reputation of OSnews.
She knows what to select for this very site. And she is very careful in selecting articles.It just happens that you don’t like something.Is a news site , you know.As for the reputation you should not worry, is better than yours anyway.No , get to work and quit browsing the net during your regular program.
This article would from its very style of writing indicate to me that the author has a haphazardly organized frame of mind.
God help his users.
If he’s committing his thoughts down and they look like this, you can only imagine his instructions to the user base.
People like you are precisely why microsoft has a strangehold on this market space. What is not properly communicated cannot be used!
Regards,
Microsoft Fan Boy
A few weeks? What are you talking about? As someone who works for an actual professional system integrator … I can set up a secure Samba file server in less time than it takes to just install 2K3 and apply all the updates from Microsoft (which my company does on a regular basis as well … we’re no OSS zealots, we’re just trying to get the job done).
Okay what about supportablity of the server once its in place. In the organazations i have work in the past once a windows file server is setup properly tasks such as granting users permissions to shares,creating new shares etc.. were transfered to helpdesk personel. I don’t see this happening with a similiar Samba solution easily.
If you’re interested, here’s a very nice pamphlet of BSD success stories, just put together by Dru Lavigne for O’Reilly.
http://linux.oreilly.com/news/bsd_ss.pdf
We have seen enough bad articles. What is enough is enough.
Why should we migrate to anything?! We’re happy with AIX, Tur64, HP-UX, IRIX and even Solaris. For all of them we have latest GNU software, free software and freeware. It’s true that software under GPL is more likley to get compiled under BSD or Linux, but hey that’s software fault as programmers are ignorant to other users/systems.
“You can integrate a Samba 3 server in a 2K or 2K3 Active Directory environment, even have it serve as an AD domain controller.”
I am as much a UNIX/Linux/Mac OS X fan as the next, but you can’t have Samba 3 act as an AD domain controller. It acts as an NT 4 domain controller, which is different. Thats not to say that a NT 4/Samba 3 domain controller can’t serve the needs of most customers. Samba 3 can act as an AD domain “member”, not an AD PDC.
“Is there any reason to replace a Windows based fileserver with a Samba fileserver? Maybe, but does it make sense when looking at the larger picture? Hardly. And would a BSD solution have any chance of seeing the light of day in any sensible IT organisation. No way. And that’s not to be mean, but that’s a sense of reality. Where are the certified professional for BSD? Where is the vendor backing? Where is the garuanteed working hardware from industry partners (and don’t even try to mention BSD Hardware Inc. from Backwater, USA. I mean HP, IBM, Dell, the usual suspects…and what about unimportant players like, oh, Getronics to actually support the whole infrastructure, including emergency recovery, etc). In other words, who do I turn to when the sh*t hits the fan and I have to get things working again by 5pm *today*? ”
OK since we are talking about UNIX here, and not just obscure distros of *BSD, you might want to consider there is Sun, IBM, Apple, etc supporting fully commercial variants of BSD, UNIX, and Linus that _do_ have consultants, _do_ have a support infrastructure, _do_ have certified hardware, and _do not_ have the problems of Windows (viruses, malware, crashing, downtime).
Why replace a Windows file server with say Mac OS X Server, or IBM supported Linux? Oh I dunno, I don’t like it when my business can’t bill hours because some stupid virus took out my fileserver, or I get a BSOD because some hack job at MS didn’t debug their stuff because they have no motivation in the market to do so…