Want to read Slashdot from your mobile device? Point it at m.slashdot.org and keep reading!

 



Forgot your password?
typodupeerror
Check out the new SourceForge HTML5 internet speed test! No Flash necessary and runs on all devices. ×

Microsoft to 'Support and Usurp' Unix 102

qedramania writes "Computerworld has a report on the latest Windows server release and their Unix strategy." From the article: "R2 is built on the Windows Server 2003 Service Pack 1 and is geared towards specific workloads such as storage management, branch office server management, as well as identity and access management. It also provides a subsystem which supports Posix applications."
This discussion has been archived. No new comments can be posted.

Microsoft to 'Support and Usurp' Unix

Comments Filter:
  • by eldavojohn ( 898314 ) * <[moc.liamg] [ta] [nhojovadle]> on Tuesday March 14, 2006 @08:50AM (#14914863) Journal
    To manage the applications, R2 will provide shells - a command language interpreter - to run Unix scripts and Telnet clients. On the tools front, Visual Studio will provide a debugger for Posix applications. These developments will make it easier for users to migrate Unix applications to Windows, said Lowe.
    That, in conjunction with enabled NFS & Unix Network Information System support, looks to me more like assimilation than usurpation. I think it's obvious that we're going to see Microsoft try to migrate Unix server applications to their server platform while at the same time trying to pluck the best parts of Unix (hopefully security!) for their own OS. Let's not kid ourselves, both sides could learn a lesson from the other in a wide variety of areas.

    I guess Slashdot's picture of Gates as a Borg [slashdot.org] is applicable more now than ever ... but, in my personal opinion, I kind of see this as a good thing.
    • I guess Slashdot's picture of Gates as a Borg [slashdot.org] is applicable more now than ever ... but, in my personal opinion, I kind of see this as a good thing.

      You mean he is assimilated in the Unix-hive, don't you? ;-)
    • by babbling ( 952366 ) on Tuesday March 14, 2006 @09:23AM (#14915003)
      It really depends on what they're up to. Their past efforts have shown that they're not too interested in being compatible with Linux/Unix, so this is suspicious.

      One reason they might be doing this is to counter free software. Currently, projects like Samba have been making good progress toward connecting Unix and Windows computers. Samba is free software. By Microsoft closing the connectivity gap themselves, they can close it with closed-source, proprietary software. This means that they can control connectivity. If they so choose, they would then be able to break it off completely at any time.

      So, this might just be a grab for power.
      • by Bloater ( 12932 ) on Tuesday March 14, 2006 @09:53AM (#14915166) Homepage Journal
        I think Microsoft is *very* interested in being compatible with Linux. If Windows can become the very best choice for heterogenous networks, then it becomes an easy decision for *any* users and *any* network. The default question when making a buying decision will be "What compatibility bug prevents us using Windows here?". If there isn't one, they'll buy Windows.

        Part of the reason for enterprises to choose Linux so far has been "It works in nearly any point in our network, so we can always just install Linux servers". Since purchase price is not a big issue for business users, the only downside to Windows would be client access licenses. If they got rid of those and bumped up the initial purchase price of Windows server systems, Linux would be hurt very, very badly.
        • by r00t ( 33219 )
          It's not the dollar value itself that is the problem. It's getting approval. Somebody has to sign off on the purchase and make it happen.

          Then there is the matter of storing and keeping track of those silly hologram cards that supposedly prove that you have valid licenses. It costs staff time to deal with that. If you screw up, and maybe even if you don't, the BSA shows up with a bunch of US Marshals (or non-US equivalent).
        • This is an excellent point. I manage a few dozen mixed servers (Windows, Linux, Solaris) that fall into one of three categories: Production applications (web, database, email, etc); Testing for the production applications; and various utility services (internal ftp, service monitoring, backup etc).

          The utility servers are without exception Linux because they can be made compatible with any or all of my other systems with very little work and with no capital cost. Currently, Windows isn't even on the radar wh
      • They don't like being compatible with Linux/Unix when we are talking about Windows to Unix.
        However they are not stupid, they must make sure people can migrate to windows nicely with the less trouble as possible and AFTER lock them in proprietary technologies.

        They want to kill Open source ? Sure but I don't think this release in their new killing machine, just a part of their global strategy. In fact, the posix subsystem has been around for a while and is beter know as Interix or Windows Service for Unix. Th
    • by IntlHarvester ( 11985 ) on Tuesday March 14, 2006 @09:34AM (#14915049) Journal
      If anything, I'm suprised it took them this long to take "UNIX" seriously.

      One of the biggest thing that's been going on in enterprise computing is the widescale dumping of UNIX/RISC in favor of Linux on cheap x86 boxes. Microsoft has almost entirely missed out on this movement. If they can make Windows into a half-decent *nix, there's certain a big growth opportunity for them.

      And while the usual crowd is suspicous of MS's motives, I'm sure there's some developers out there excited about Microsoft embracing* a non-proprietary "industry standard" API. It's a big step for them. No longer would you need special Windows ports of software like Apache and Postgres --- in theory, you could just "make install".

      * word used with caution
    • by ILikeRed ( 141848 ) on Tuesday March 14, 2006 @11:20AM (#14915834) Journal

      To manage the applications, R2 will provide shells - a command language interpreter - to run Unix scripts and Telnet clients.

      So typical of Microsoft - it's 2006, and to compete with Linux they start offering Telnet clients rather than something actually useful and secure like ssh. I can picture the sales calls and interviews right now, "well, they insisted they wanted Linux compatibility, it's not our fault that Linux telnet is so insecure, if only you had done your implementations the Windows way."


      • Yeah, this is probably a lot like "POSIX compliance" for Windows NT. It was there to meet bullet points on requirements documents, rather than to actually provide the functionality.

        If Windows had really implemented POSIX, then why is there Services For UNIX, Cygwin, and the MKS suite?

        In the past decade+ that I've been exposed to Windows, never have I once had the impression Microsoft was really interested in interoperability or compatibility, beyond some token marketing-oriented effort.
    • I guess Slashdot's picture of Gates as a Borg is applicable more now than ever

      Actually, I wasn't thinking Borg. My first thought was, "All your bases are belong to us."

      But is basically an acknowledgement that *nix is providing things they aren't and that there are benefits to it.
    • ...but I still don't see the big differentiator here that'll convince a UNIX shop to adopt Windows over the latest iteration of their existing platform.

      This isn't a really huge move actually--it is just more of the same "bundling" stuff that Microsoft has done with its OSes forever (applets in Win 3.0 is where it started and now we have Media Player, IE, firewall, etc). Microsoft has finally seen how successful projects like Cygwin have eaten into its SFU market space, and relatively speaking SFU has been
    • If they are supporting telnet, they are obviously not very interested in the security angle. They'd be doing ssh if they were.
  • by tsa ( 15680 ) on Tuesday March 14, 2006 @08:50AM (#14914864) Homepage
    If you can't beat them, join them!
  • i mean posix has been long available under NT, nothing new under the sun. Microsoft allready had als a fileserver update to have size limits on folders. In the end this is not unix specific it was user environments require.

    There are a lot of similarities ofcourse however often they are result of standardisation, or the result of new technoligies, if you think about a SAN environments then folder based limits makes sence. I see more and more SAN's connected to MS networks.
  • Oh, great... (Score:3, Interesting)

    by mrchaotica ( 681592 ) on Tuesday March 14, 2006 @08:56AM (#14914886)
    If it were anyone other than Microsoft I'd be happy about POSIX support, but you just know they're going to make it "MS-POSIX" or "POSIX++" or something stupid instead, and cause more incompatibility than they fix.
    • I bet they'll call it POSIX# or POSIX.NET...
    • Re:Oh, great... (Score:5, Informative)

      by jb.hl.com ( 782137 ) <joe&joe-baldwin,net> on Tuesday March 14, 2006 @09:12AM (#14914950) Homepage Journal
      Microsoft has had a POSIX subsystem for ages. It's called Windows Services for UNIX, and it works quite nicely. It's not a new thing.
      • The Docs for it kinda suck, but when you get it figured out, It does the job. I've been mounting NFS shares on my Winders box using SFU for quite some time. as an added bonus, ls works too!
      • I've used SFU. It's like going back to the bad old days of Interactive Unix for the 386, around the early 1990s. It should in theory be much better than cygwin because it's an actual subsystem, but in practice I've found cygwin works far better.
    • POSICKS ?
      POSIX# ?
      POSIX.NET ?
      MoSftIX ?
      ME-NIX / YOU-NIX (from Win-ME)
      which leads of course to the British slang ... WE-NICK [answers.com]...which is of course Billy's real talent.
    • Everybody who replied to the parent, hurry up and trademark these phrases. We don't want them to actually be used by Microsoft. So make sure you don't let them use it, even with royalties.
    • SFU is more like POSIX--. It's about buzzword compliance, not about being useful.
  • Yawn. (Score:4, Insightful)

    by Vengeance ( 46019 ) on Tuesday March 14, 2006 @08:56AM (#14914887)
    Will they get more than an '80% POSIX complaint' OS out of this effort?

    And does anyone who uses a real UNIX actually care?
  • by pandrijeczko ( 588093 ) on Tuesday March 14, 2006 @09:02AM (#14914909)
    At a recent media briefing in Singapore, David Lowe, senior product manager, Windows Server, Microsoft, cited interoperability with Unix as one of the key features of Windows Server 2003 R2.

    So how come Jeremy Allinson and the other SAMBA guys have such a problem getting technical details out of Microsoft about the inner workings of SMB for their product that allows "Windows interoperability with Unix"???

    • by Azarael ( 896715 ) on Tuesday March 14, 2006 @09:35AM (#14915055) Homepage
      By interoperable they mean that Windows software will be able to make use of Unix software without having to give anything back in return. Otherwise known as a 'One way street'.
    • by lkcl ( 517947 ) <lkcl@lkcl.net> on Tuesday March 14, 2006 @09:41AM (#14915098) Homepage
      the POSIX subsystem, like all microsoft "announced products", reappeared coincidentally at around the same time that opennt.com, which provided a full third party POSIX subsystem, disappeared.

      anyone may, if you have sufficient information on how the NT kernel works or are prepared to reverse-engineer it (like the ReactOS guys are doing), write their own subsystem. there are THREE types THAT I KNOW OF: OS/2, Win32 and POSIX. okay, maybe there are four now - win64.

      having a POSIX subsystem sit on top of the NT kernel, which is a microkernel based on the Mach microkernel, is NOT the same as having fast and direct access to the NT kernel functions.

      and the reason why the samba guys have such difficulty getting information is because there either ISN'T any (it's all in the code) or there's too much!

      the only reason why the CIFS documentation effort was initiated by microsoft is because the original people who worked on it (having embraced-and-sensibly-extended the IBM Lanman SMB spec and also the X-Open SMB spec), having retired with their stock options up to millions, left no clues as to how this HORRIBLY complex code worked.

      it was therefore ESSENTIAL that they get it documented.

      the first time they released cifsbrow.txt, in 1997, because i'd just spent five months network-reverse-engineering the network neighbourhood and WINS server code, i spent a WEEK throwing email messages at them, explaining various inconsistencies, helping them improve the documentation they'd created. it takes TWO YEARS to correctly implement the network neighbourhood. it's a FULL peer-to-peer registration and management system, very robust, very complex, _extremely_ good, and people have xxxx-all idea of quite how useful it is ("oh, it's netbios - switch that xxxx off")

      after the first CIFS conference, andrew, jeremy and i hung around for an extra day: i got to meet the guy responsible for the network neighbourhood, and spent a good couple of hours drumming into him the things that had been forgotten since the email flurry - there's nothing like meeting someone face-to-face to explain stuff, as you well know.

      so.

      i'd say that the reason proper documentation doesn't "exist" is partially deliberate, and partially it's your average development/management incompetence.

      l.
      • Network Neighbourhood and Microsoft's vision of domains and workgroups actually appeals to me on paper. In practice, its administrative hell, but it looks great.

        I've implemented successful workgroups & domains since switching to Samba for all the heavy lifting and been much happier than on Windows.

        Thanks for the work.
      • it's a FULL peer-to-peer registration and management system, very robust, very complex, _extremely_ good, and people have xxxx-all idea of quite how useful it is

        You left out "incredibly chatty" and "hopelessly lag-sensitive", two attributes that make CIFS-based operations over an extended WAN environment painful at best. Ask anyone who's tried to drag'n'drop a few thousand files onto a network share located on the opposite coast from behind a half-ISDN connection. For added fun, try working with a Visua

        • Not to sound like a curmudgeon, but isn't just about everything over an extended WAN, except perhaps BITNET or UUCP batch transactions (you know, pre-old school stuff), hopelessly lag sensitive and extremely painful at best?

          At some point, it's hard to beat the bandwidth of dropping a CD (or DVD or tape) in the mail, even though the latency sucks.

          Visual SourceSafe? I'd rather give myself paper cuts over my body and then roll around in a briny vineagar solution, perhaps with a bunch of cayenne pepper sauce th
          • If you're using Windows Explorer, SMB is extremely glacial compared to FTP or HTTP/DAV/Web Folders in the same environment -- over a 128K ADSL link, which would have been blazing in the bitnet days.

            Admittedly it's better from the CMD line, so that's not entirely the protocol's fault. But it is slow -- probably a ton of error-checking and authentication overhead.
          • Not to sound like a curmudgeon, but isn't just about everything over an extended WAN, except perhaps BITNET or UUCP batch transactions (you know, pre-old school stuff), hopelessly lag sensitive and extremely painful at best?

            Can't speak for BITNET (never used that one), but UUCP was also horribly lag-sensitive, given that it had a one-frame window (which is to say, no window at all). Some folks may remember the venerable Telebit Trailblazer modem's UUCP spoofing feature (it faked the ACKs locally to cut do

        • You left out "incredibly chatty" and "hopelessly lag-sensitive", two attributes that make CIFS-based operations over an extended WAN environment painful at best.

          So, how are you going getting that screw out with your hammer ?

      • Ian Nandhra of Lasermoon in UK - who sold and promoted Linux-FT - created OpenNT to move Linux POSIX code and utilities onto a Windows environment. OpenNT got threatened with a law suit - changed to Integrix - and were bought out by Microsoft. Code was later released as SFU by Microsoft which may well be the direct descendant of Linux-FT and Ian's work. Ian put the Linux kernel through POSIX compliance - see dmesg - but has now vanished off the planet. I'd like to talk to him at some point about FT and the
  • by Anonymous Coward on Tuesday March 14, 2006 @09:03AM (#14914916)
    Why do article submitters assume us UNIX guys know all the latest buzzwords and upstart companies in computing? For the uninitiated, Microsoft [microsoft.com] are a software publisher who sell a toy operating system called Windows [microsoft.com].
  • by Anonymous Coward on Tuesday March 14, 2006 @09:07AM (#14914929)
    How many programs were developed for UNIX (or Linux, or ...)? For example, Apache? Now runs on Windows. Postgresql? Now runs on Windows. There is a LARGE amount of formerly UNIX-only software, much of which is open source, that can now run on Windows. Microsoft is no dummy. They too, just like SCO, can leverage this software.

    Open Source is a double edged sword -- it gives you a fantastic advantage, but at the same time, your competitors are free to use your software, your IP, your efforts. One hopes that the benefits outweigh the advantages to your competition.

    The real strengths of Open Source are leveraging development and testing all over the world (lower product costs, time to market, code reuse, etc.), much lower marketing and sales costs (Internet distribution), and better quality (many eyes make all bugs shallow).
    • > How many programs were developed for UNIX (or Linux, or ...)? For example, Apache? Now runs on Windows. Postgresql? Now runs on Windows. There is a LARGE amount of formerly UNIX-only software, much of which is open source, that can now run on Windows. [...] Open Source is a double edged sword -- it gives you a fantastic advantage, but at the same time, your competitors are free to use your software, your IP, your efforts. One hopes that the benefits outweigh the advantages to your competition.

      If not f
    • by CastrTroy ( 595695 ) on Tuesday March 14, 2006 @09:43AM (#14915112) Homepage
      The problem is, is that a lot of the unix software although ported to windows, doesn't run as well on windows. (i think) Apache starts a new process for each request by calling fork(). This works good in Unix where it doesn't take a whole lot of resources ( under a million cycles) to start a new process. However, just starting a process on windows takes 5 million cycles. Meaning unless the implementation of apache on windows, or windows itself is changed, then running apache on windows will always be inferior. I imagine the same problem exists on many other servers that have been ported to windows. It's nice for people to experiment with these tools without switching their OS, but really they should be switching the OS if they want the best experience possible.
      • Meaning unless the implementation of apache on windows, or windows itself is changed, then running apache on windows will always be inferior

        In fact, Apache on Windows does use threads, and not processes. From http://httpd.apache.org/docs/2.0/platform/windows . html [apache.org]:

        Because Apache for Windows is multithreaded, it does not use a separate process for each request, as Apache does on Unix. Instead there are usually only two Apache processes running: a parent process, and a child which handles the requests. Wit

        • So in windows it uses threads and unix it uses a separate process. Is there any advantage to running the separate process, or could the just switch the unix implementation to threads, to get even better performance? There must be a reason why the unix version is still starting a new process, which is always more costly than just starting a new thread.
          • Well, first of all the UNIX version can be configured to use threads. There are several other ways to run Apache.

            Second of all, UNIX was dragged kicking and screaming into supporting threads. Lots of UNIX-like systems are still very bad about thread support. Thread creation is often slower than process creation. It was MUCH slower on Linux 2.4.

            Third, threads share memory. This can be a curse. It is fundamentally less reliable. If the server has a bug, it will crash if running in threaded mode. A forking ser
      • Apache does not fork a new process for each request in Windows, it actually just creates a new thread. (Source [apache.org]) Not that I would use it in Windows anyways though.
    • Because a great deal of the programs you mention are generally ported to Windows by the SAME DEVELOPERS. Microsoft isn't leveraging anything, quite the opposite. FOSS tends to undercut MS's flagship products on the same platform. Microsoft CANNOT use GPL-compatible sourcecode, their employees are forbidden to even look at it. They prefer to take BSD source, which is no great benefit, since everyone else can too.
  • MS has had a site for Unix migration for a long time. Resources for UNIX Professionals [microsoft.com] provides various takes on migration. Plus, as has been noted, Posix has been part of NT at least since version 4 and perhaps even 3.51 although I'm too lazy to look it up.
  • This is news? (Score:1, Informative)

    by millerjl ( 126046 )
    This isn't news. There has been POSIX Support in NT4 and win2k forever (so it seems), and Windows 2003 already can do NIS if you know what you are doing with schemas and the Services for Unix. The only thing "new" would be unix shells native to the OS... but this can be done effectively now with other packages like cygwin [cygwin.com] or MinGW [mingw.org].

    • It is news because it is now built into Windows 2003 Server.

      The Unix features will also be built into certain versions of Vista later this year.
    • It is if it's fully (or almost fully) POSIX complient. Microsoft promised us NT would run win16/32 apps, OS/2, apps, and UNIX/POSIX apps back before NT came out. It turned out they did a real half-assed job of POSIX support. Yes a very few POSIX apps will run on NT/2k/2003, but most won't. To get most of them to run you needed to install cygwin, etc, or it's just not going to happen.

      *Real* native POSIX support IS news.

      • I never said it was good POSIX support. Yes, it's Microsoft's limited version of POSIX, whose hooks I promptly delete/disable from any version of OS I install that has it included. Neither the above article nor MSoft's web site on R2 states that it will be Real or Full POSIX support. MSoft just talks about Interoperability Components [microsoft.com], which already exist, and thus isn't real news. If they are putting in Real POSIX support, they are not publicizing it well or supporting the contention that it is a fully com
  • Microsoft Xenix (Score:3, Informative)

    by Ratbert42 ( 452340 ) on Tuesday March 14, 2006 @09:23AM (#14915000)
    Microsoft's been doing Unix [wikipedia.org] since Linus an elementary school kid playing with his Vic-20. It was the first Unix I used, running on Tandy hardware.
  • by gonar ( 78767 ) <sparkalicious@verizon. n e t> on Tuesday March 14, 2006 @09:25AM (#14915017) Homepage
    and it has always sUxx0rd. incomplete, poorly implemented, not really POSIX.

    are they saying that they are doing it right now, or just pretending what is old is new?
    • I recall my OS prof saying NT was designed by the VMS folks to have POSIX support from the start, but market pressures suppressed using that functionality(MS wanted people to buy their win32 software rather than run sendmail/bind/etc on NT machines). Ah, here we go:

      Microsoft hired a group of developers from Digital Equipment Corporation led by Dave Cutler to build Windows NT, and many elements of the design reflect earlier DEC experience with VMS and RSX-11. The operating system was designed to run on mu

  • Microsoft has been doing things for unix for a while. Microsoft actually had a big unix side. They eventually sold a big portion of it to a small company that i used to work for called proginet . I was amazed at all the microsoft stuff they had. They had some microsoft applications that I didnt even know existed .

    I find it funny that now they are trying to go back to unix stuff.
  • by NZheretic ( 23872 ) on Tuesday March 14, 2006 @09:36AM (#14915065) Homepage Journal
    Windows Server 2003 R2's Unix interop feature is derived from Microsoft's Services For Unix (SFU) which pulled a lot of source code from OpenBSD compiled by and packaged with GNU GCC.

    For a full history of NT, Interix and SFU, see Should that not be GNU/Microsoft SFU? [oreillynet.com]

    Many Microsoft users will be running a modified version of the GNU system every day, without realizing it.
    • Windows Server 2003 R2's Unix interop feature is derived from Microsoft's Services For Unix (SFU) which pulled a lot of source code from OpenBSD compiled by and packaged with GNU GCC.

      What's your source for compiled with GCC? I'd have thought they'd use their own compilers through the /bin/cc wrapper.
    • by Anonymous Coward
      For a full history of NT, Interix and SFU, see Should that not be GNU/Microsoft SFU?

      Oh, ahahaha. *You* wrote that. Or, rather, you plagarised it from Stallman's GNU/Linux essay and search/replaced it to be GNU/SFU. Are there any real facts in there at all?
    • C is C is C (or rather, it ought to be, because that was the whole point of C).

      The OpenBSD people pride themselves on being correct. I would expect that they would have written tight, neat, standards complient C. The BSD libraries are not ripped off from GNU. So, MS should have no problem compiling and linking a non-GPL version of the BSD code. Beside just using GCC would not emcumber you if you link against your own libs and headers, would it? How do you think the BSD people do it and avoid the GPL whi
  • Microsoft lost the golden age of UNIX. It implemented a half compatible system and lied to people telling them it was compatible. So most people kept UNIX.

    Now, UNIX is dying anyway. Linux and BSD (mostly because of GNU tools) are each day less POSIX compilant, and the programs are using their extensions (that are quite usefull). Even more important, proprietary UNIX are almost dead aready, why does MS thinks that it will be sucessfull introducing a new proprietary UNIX now? Why does it think that people wi

    • Even more important, proprietary UNIX are almost dead aready, why does MS thinks that it will be sucessfull introducing a new proprietary UNIX now? Why does it think that people will buy that UNIX.

      They're not "introducing a new, proprietry UNIX", they're offering a migration path from existing unix systems to Windows.

  • I honestly don't see how they find the time do all that and spew all the FUD.
  • A few years ago, some IT shops might drop Unix because it doesn't support Windows properly. Now, some IT shops might drop Windows because it doesn't support Unix properly.

    I see that happening more in a server farm than desktops, but it's a start.

  • Posix, Shmosix. I want libraries. I want REAL built-in UNIX apps, not MS work alikes. When MS drops DLLs and The Registry and learns to play nice with others, wake me up.
  • Small example - symbolic links are not in the POSIX standard, even though every Unix since BSD 4.x (4.2? I forget...) has had them. Nobody's Unix compatibility system for Windows does symbolic links, because they're not a good fit for the Windows file system - the closest you can come is linking to a directory, not a file.

    Windows has a POSIX-compliant subsystem, and has had it for years, but it doesn't do symbolic links, not completely. How many non-trivial Unix applications don't use symbolic links, eith
  • ...in the guise of support. Usurp as in burn cycles, throw the DB into a "hurry up and wait" state, and have non-programmers cobbling together their own "applications" on the client side in blissful ignorance of petty annoyances like referential integrity.

    I just can't wait to see what MS does next to "support and usurp" Unix.

  • Seriously, it's not like I need Microsoft to do anything I want with computers, so why add the bloat?
  • I remember reading that one of NT's big features was going to be the Windows, OS/2, and POSIX subsystems on top of a lightweight kernel. (Sounds similar to MVS and OS/400, two extremely powerful virtualized operating systems).

Swap read error. You lose your mind.

Working...