“In recent years, Linux has proven itself a credible alternative server operating system, and application support has also improved greatly. These days, the question is not whether you can use Linux, but where you can best use it. Is there more to Linux than Apache and file and print serving? If you’ve spent any time hanging around Linux-friendly discussion sites such as Slashdot in the past few years, you could be forgiven for thinking that Linux is still poised to take over the world. Reality is not quite as kind.” Read the article at TechUpdate.
Whose reality?
TechUpdate’s. This is their text.
The first lines of this article, as quoted on OSNews, strongly woke my interest, more than the average article.
Is there more to GNU/Linux than Apache and file and print serving?
This is the question I keep asking my self all the time. First of all, let me state that I’m not a *nix specialist. I admit, I’m a Microsoftie. I grew up with it, made it the primary knowledge for my job etc. Within the last few years I started playing with GNU/Linux and discovered the strong basis of it. If GNU/Linux and other OSS could replace the tools we use daily, I would be happier. Please note – and this is important – I’m talking about the business and markets I work in: very small companies with mostly 20-30 employees and only sometimes up to maybe a 100. Companies who can’t afford a full-time IT staff, and who hire me and my colleagues for a couple of hours/days in the week to keep things running.
Yes, we sometimes use GNU/Linux. A firewall, an intranetserver, a dns or mailrelay, just to name the typical few. The problem is those mostly are stand-alone servers, hard to integrate with an existing Windows network. Now before you start to flame me, please try to understand the point I try to make. Yes you can serve files, you can host a domain with GNU/Linux – but that is just so much harder to do than with a genuine Microsoft product. Face it, you have to integrate all pieces of software yourself before you could have the same possibilities. Every article I read mentions Samba making it possible to server files on Windows networks. They never mention that they’re talking about folder sharing, not that file-system security is something completely different. Apologise me if I’m wrong bu AFAIK but I don’t see how you could manage the same granularity of file security with a Samba server.
I’m not trying to bash on Samba and other OSS. I just love them for existing and I praise the guys who make them possible. I’m not saying that it’s bad software. My point is: people who want to compare Microsoft products with GNU/Linux and OSS, should pay some more attention to the very technical differences. I don’t say you should use MS because OSS can’t do the job. I’m saying that despite all the good words are spoken about GNU/Linux, there isn’t any distro who provides one single product which could exactly replace a Windows server.
Should it? I’m not saying every GNU/Linux product should. I’m saying the product which *might* be a killer application to replace Windows on the server side should.
Other examples? (And – again – I’m mean examples of bad articles. Articles who try to say that OSS provides alternatives – which they do – but forget to say that it ain’t that simple.)
Active Directory you say? Sure, OSS provides a bunch of LDAP servers which do the same job. – Can you use them to integrate them with Windows software? – Yeah sure, you can! – OK, which product do I need? – Well you need this *LDAP, that module, that client from Samba, and maybe this and that. – Ough – that’s a pain to setup and configure. My customer just pays me one day to set this up.
Now you all might understand that this article just turned out to be an average one…
Note that I don’t even bother to start talking about Linux on the desktop. I am happy with the recent M9 & RH8 and other evolutions. But it’s just not enough for me and certainly not for my customers. Besides, the desktop side of the story isn’t the most important one. Server side is. Ever heard of CAL’s? Server related licences are the toughest to pay. And even this is so relative. Sure Linux is free as in beer. But What is say 5k-10k$/โฌ for a couple of licenses compared to the two to three times as much for buying the hardware? And I even didn’t count the yearly budget my customer pays to hire me.
To finish, a bunch of questions I would like to see a good answer to. I know there are good answers. But they are just not good enough for me, my boss AND my customers who want more than that.
– How do I really integrate Linux into my Windows network?
– How could I replace my ISA server with SQUID, without requiring my users to retype their Windows logon credentials?
– How do I setup a VPN server, with which my users can authenticate with theirs Windows account, using the standard Microsoft dial-up client, AND without a hassle to setup a transparent name-resolution system!
– Please tell me about a Mail server who [i]transparently[i] can replace Exchange.
– How can I have a central AND single-sign-on authentication for my Windows desktops without using Windows servers?
– …
It might be time to start a portal “Real Linux alternatives for Microsoft Professionals”? Not that OSNews ain’t good, it just isn’t the same scope ๐
I hope this comment didn’t offend anyone. This is just about stuff I and most people thinks when they work intensively with Microsoft products. And there are a whole lot of us out there.
Wow. You hit the name right on the head. You’ve pointed out the main problems with Linux servers. Not so much compatibility, but hassle. For the small business, it’s just not worth it. Techs (although Linux requires less maintnence, it does require more skilled techs, and most tech-savvy people who would also do something else at the small business can configure Win2K Server–at least the basics), downtime, compatibility. Ugh. Hassle.
Windows servers for Windows networks, *nix servers for *nix networks. It’s just easier. I think I should copyright that…
It’s Just Easier (r)
File Servers:
XFS supports Access Control Lists (and ACL patches are available for JFS and ext2/ext3), allowing the same file security granularity as Windows. Unfortunately support for this has yet to be integrated into any of the popular filemanagers, and AFAIK nobody has developed a GUI for this at all. The command line tools are fun for geeks but probably not quite what you’re looking for…
IIRC Mandrake takes advantage of this in their tweaked version of Samba, supporting ACL’s on XFS partitions.
Personally, I wouldn’t bother with Samba for file serving, I’d use NFS (or better yet AFS – much more secure than NFS or Windows file sharing) drivers on the Windows clients. Kerberos could be used to centralize and unify logon and idenfication services. NFS support can be found among the MS Windows Tools for Unix, perhaps AFS and Kerberos too? Anybody know of a good CUPS client for Windows?
Linux/*BSD/Unix use open networking and security protocols; therefore, its easier to make Windows Linux compatible than vice-versa.
Exchange/Active Directory:
I believe there are commercial equivalents made for Linux, *BSD and Unix servers. Caldera/SCO and Novell have directory software. Lotus Notes/Domino offer even more groupware functions than exchange, and HP developed and exchange-like platform others have licensed (I forget the name). At the moment these are ‘enterprise’ solutions, which would probably cost more than Windows for your small business clients.
Conclusions:
The tools you need are possibile, and much of the foundation is in place. With IBM, HP/Compaq, Sun, etc. supporting Linux, these tools will most likely be available soon.
–Raging Dragon
>File Servers:
>
>XFS supports Access Control Lists (and ACL patches are available for JFS and ext2/ext3), allowing >the same file security granularity as Windows. Unfortunately support for this has yet to be >integrated into any of the popular filemanagers, and AFAIK nobody has developed a GUI for this at >all. The command line tools are fun for geeks but probably not quite what you’re looking for…
Baah, ACL’s are for kids. You want a file? You really want that file? Send me an email. Prey it doesn’t bounce. When I get it, if I’m not too busy I’ll give you that file. heh heh (Kidding) Seriously, ACL’s are great but unix permissions will work fine as well in a small environment provided you are properly restricting your network. IE: Unused ports are off, Live network addresses are restricted by IP addresses. Unused IP addresses are in hosts.deny. The network is monitored for MAC address changes, and when detected IP addresses are added to hosts.deny. When assigning IP addresses never ever assign them in series. IE: .1 .2 .3 is BAD, use .50 .61 .72 etc until you have no other alternative. Internet access -> Firewall/Linux (Linux/web,Linux/DNS)-> Firewall/FreeBSD -> Network
Enough rambling, I’m sure you get the point. haha
>IIRC Mandrake takes advantage of this in their tweaked version of Samba, supporting ACL’s on XFS >partitions.
>Personally, I wouldn’t bother with Samba for file serving, I’d use NFS (or better yet AFS – much >more secure than NFS or Windows file sharing) drivers on the Windows clients. Kerberos could be
Samba is good. You can still use kerberos, life is good. Me, I’m bold. I’ll use NIS to replicate passwords between servers, and restrict network access. (Heh send me an email if you can’t get on the network. I’ll turn your port on, and assign an IP address as soon as I see that message.) <VEG> ๐
>used to centralize and unify logon and idenfication services. NFS support can be found among the >MS Windows Tools for Unix, perhaps AFS and Kerberos too? Anybody know of a good CUPS client for >Windows?
Windows -> Samba -> Printer, has been good to me. Unified logons are no big deal, remote connections with your *WINDOWS PASSWORD*? WOW, I can’t believe you said that out loud, much less in public :-X If you are using *WINDOWS PASSWORDS* for any form of RAS, you are dumb enough to follow my next instructions. Go to you managers office, and tell him to fire you. Seriously, if you aren’t using some form of secure id access you will be (or have been) hacked, it’s just a matter of time.
>Linux/*BSD/Unix use open networking and security protocols; therefore, its easier to make Windows >Linux compatible than vice-versa.
Yes.
>Exchange/Active Directory:
>
>I believe there are commercial equivalents made for Linux, *BSD and Unix servers. Caldera/SCO and >Novell have directory software. Lotus Notes/Domino offer even more groupware functions than >exchange, and HP developed and exchange-like platform others have licensed (I forget the name). >At the moment these are ‘enterprise’ solutions, which would probably cost more than Windows for >your small business clients.
Insite, Communigate Pro, OpenMail, Domino, or something web based.
>Conclusions:
>
>The tools you need are possibile, and much of the foundation is in place. With IBM, HP/Compaq, >Sun, etc. supporting Linux, these tools will most likely be available soon.
>–Raging Dragon
It’s all there, and has been for years.
I’m saying that despite all the good words are spoken about GNU/Linux, there isn’t any distro who provides one single product which could exactly replace a Windows server.
Well, of course. If you want something that acts exactly as Windows, you’d better keep Windows, after all. There’s nothing wrong in that (but yes it’s expensive ;-)).
Linux and free software is not about mimicking Windows nor proprietary software. It is about doing (perhaps different) things in perhaps different ways.
So let me turn your sentences the other way around :
– there isn’t any single Windows edition which could exactly replace a Linux distro
– the problem is those mostly are stand-alone Windows servers, hard to integrate with an existing GNU/Linux network.
– etc.
If you’re thinking about a way to simply replace Windows, then maybe you’re looking at free software the wrong way.
Regards
The poster makes the valid point, and I think it’s a sign of the _success_ of linux, not a sign of failure, that what windows admins have come to expect is a pre-packaged environment that will be a drop-in replacement for all the “basic” windows-network services.
They (I guess, really, “we”–I’m one of them about a quarter of the time) figure that if it isn’t easier to upgrade their server(s) to linux than it is to the next version of windows, then it does not represent the “lowest cost” upgrade in their eyes or in the eyes of their customers. This is valid, if short-sighted.
If we want these support professionals to go with linux solutions, then we need to jump this hurdle, not really in terms of technology any more, but in terms of packaging. Luckily, once a need becomes visible, so far the oss community has eventually filled it successfully.
Eventually (and hopefully sooner rather than later) there will be a single distribution created that will provide a pre-built, drop-in replacement for all of the basic services the poster mentions, with simple, web or gui admin clients.
It’s clear that a small team of experts could create this today, if they chose to. Back up your data, drop the cd into the windows server, and reboot. No fuss, no muss. Better yet, a full-scale migration tool-suite that would move your windows/AD users, groups and file-security to a new server, transparently. It’s my hope that this is what distros versions like RH Adv Svr or top-end SuSe are shooting at.
It’s coming. I’ve got six or seven customer networks I’d deploy it to tomorrow. I just wish it was here now. Keeping Exchange and Outlook running is killing my profit margins.
Joe
Well, of course. If you want something that acts exactly as Windows, you’d better keep Windows, after all. There’s nothing wrong in that (but yes it’s expensive ;-)).
I totally agree. The point I tried to make is that I’m getting frustrated by alle the artciles who try to convince people that GNU/Linux software is an alternative for Microsoft, but forget to mention the small prints.
If you’re thinking about a way to simply replace Windows, then maybe you’re looking at free software the wrong way.
Maybe I am. But being someone who works in environments where Microsoft is the only software supplier, I have to be realistic. If I want to be able to use other tools, I need a replacement. A migration plan. This is not about designing a new infrastructure from scratch.
– be a single distribution created that will provide a pre-built, drop-in replacement
I have a dream ๐
Although possible from a pure technical point of view, I do not believe the OSS community is ready enough to provide such a solution. This is not only about just ‘making a distro’, but also about trial and error to make the right interfaces. It will need a certain time to develop that too, just like there is a lot of time needed to provide a nice look-and-feel for X, which is just only starting now.