Slashdot Mirror

← Back to Stories (view on slashdot.org)

Mounting Virtual Drives as Physical Drives in Windows?

Posted by Cliff on from the but-what-about-the-DMCA dept.

Bombcar asks: "Samba 3.0 is an excellent CIFS server, but there are some limitations. For one, you can get a networked mapped drive, but some programs (Oracle, Exchange) refuse to run on a mapped drive, but only on local drives. I know there are some closed source (read: expensive) drivers that allow a SMB share to appear to be a physical disk. Is there any equivalent in the OSS community? What I want to be able to do is mount a share from a Linux server under Windows 2000 and have it appear as if it were a local disk. This will allow many programs that refuse (for what ever reason) to use anything but physical drives to access the network." Might such software be seen as a 'circumvention device' as specified by the DMCA? The submittor mentions that there are companies in this market already, but that doesn't mean that it will remain safe.

102 comments

  1. Think Hardware by unixbum · · Score: 2, Interesting

    That would be a cool hardware device (a ide linked network card)

    1. Re:Think Hardware by sydlexic · · Score: 1

      I thought of that a long time ago as a solution for adding network storage to TiVo. It's still intriguing.

    2. Re:Think Hardware by Anonymous Coward · · Score: 0

      IDE is for old-timers. What about Serial ATA?

      Just imagine the controversy involved in a SATANetworking project!

    3. Re:Think Hardware by jackb_guppy · · Score: 1

      Would be good, I tried to have one created.

      Goal was two fold...
      1) Create a card that allow for full remote administration with full screen, keyboard, mouse ability. (KVM / PC-Anywhere / VNC)

      2) Allow the local PC's CD-Rom / Other drives be available for remote system. (Remote thinks it has CD-Rom available at boot)

      The card would have a NIC coming out of the back with SVGA, Keyboard, Mouse Connecors so local admin could be made it there was service issue.

      IBM new blades boxes system to have some like this built in so the blade enclosure has the connections and the CD-Rom built-in and you assign it on the fly to which blade you want to use it now.

  2. Why? by DA-MAN · · Score: 3, Interesting

    I'm afraid I can't answer your question, but I have a question of my own (and no I am not trolling).

    1) Why run Oracle or Exchange on 2000 stored on a Linux Samba Server? Having a physical drive would be so much better in that losing that disc during a network issue or what not would cripple your e-mail or database server.

    2) Why not run Oracle on Linux or Samsung's OpenMail on Linux instead of Oracle/Exchange and not bother with the headaches associated with creating a physical drive from a network map on a different architecture?

    3) Is there really a situation where this sort of thing would actually be useful or nessecary?

    --
    Can I get an eye poke?
    Dog House Forum
    1. Re:Why? by Gadzinka · · Score: 3, Informative
      3) Is there really a situation where this sort of thing would actually be useful or nessecary?


      Yes, there is. I've found that several games refuse to run from network drive. This pisses me off.

      I've got three computers at home:
      1. router/fileserver, runs linux and has very large /home exported via NFS and SMB so we both have networked home and media directories.
      2. my workstation, runs Debian+KDE 95% of the time, but sometimes I run Windows XP to play games
      3. my wife's workstation; also runs Debian+KDE and sometimes Windows for games.

      This way:
      • both workstation have minimal disks
      • all the disks are inside the fserver, mounted via LVM (so they apear as single partition capable of holding file of the size of whole partition)
      • my wife can freely log on to my computer to watch her video files on TV (my comp has TVout connected to TV) or listen to her music while I'm away (my comp has better sound ;)

      But I cannot install some games on the network drive in Windows so it forces me either to remove them before installing the new one, or buying more storage for my workstation (ridiculous when I have hundreds of GBs freely accessible via 100MB network).

      Robert
      --
      Bastard Operator From 193.219.28.162
    2. Re:Why? by TiggsPanther · · Score: 1
      3) Is there really a situation where this sort of thing would actually be useful or nessecary?

      Well, there are certainly programs which seem to work perfectly on local drive, and bomb out if you try to either install them on a network, or store some of the files on a network.

      Then again, maybe it's just that I'm doing something wrong, but Mavis Beacon 9 crashes out every time when I tell it to look to the Network for the User Save Files.
      But some sort of hack to stop the OS from telling the difference between a local drive and a mapped network-share would be an absolute godsend.

      Tiggs
      --
      Tiggs
      "120 chars should be enough for everyone..."
    3. Re:Why? by really? · · Score: 1

      1. make an ISO of your CD using your preffered method.
      2.download and install "daemon tools"
      3.mount previouly made ISO
      4. Enjoy life!

      --

      "Consistency is contrary to nature, contrary to life. The only completely consistent people are the dead." A. Huxley
  3. Locking by Samus · · Score: 5, Insightful

    One reason that those server programs refuse to run from network drives is locking. Last I read there wasn't near as fine grained locking available from a network drive as from a local disk. Plus the performance seems like it would just be awful. Can you imagine how long a query would take on a 1 gig table that had to do a table scan? Yuck.

    --
    In Republican America phones tap you.
    1. Re:Locking by Curtman · · Score: 4, Interesting

      So why is this type of thing perfectly acceptable on Unix boxen using NFS? Does NFS not fit your definition of 'network drive'? I would also have to ask why on earth anyone would use Oracle over CIFS, when they could just run Oracle on the Linux box to begin with? I call troll on the entire story.

    2. Re:Locking by Anonymous Coward · · Score: 2, Informative

      > So why is this type of thing perfectly acceptable on Unix boxen using NFS?

      This type of thing is not perfectly acceptable using NFS. For the program that was mentioned in the summary (Oracle), they only support a very small number of remote file options. The Oracle Storage Compatibility Program tests whether remote storage implementations will work without trashing your data in Oracle. Most NFS implementations fail it (including big names like Sun and IBM).

    3. Re:Locking by Anonymous Coward · · Score: 0

      So why is this type of thing perfectly acceptable on Unix boxen using NFS?

      It's BOXES you unwashed heathen. Boxen is not a word, it's a term used by filthy Debian zealots when they refer to each others assholes.

      I call troll on the entire story.

      I call the cops when I see you turning tricks on the corner.

    4. Re:Locking by Anonymous Coward · · Score: 0
      Boxen is not a word


      Oh?
    5. Re:Locking by drix · · Score: 2, Informative

      Ooh you gotta love OSCP. Here's I think one of the "cheaper" Oracle-certified NFS implementations out there, the NAS8000. Only $39,995 MSRP for a 4U cabinet with a whopping 0mb of included data storage. Compare with the free-as-in-speech NFS implementation that everyone conflates with "the" NFS, as if there were only one, proper noun, singular, and it were just free for the taking. Hah.

      Heh all these spare-bedroom Linux hackers these days seem to forget there's like, entire different universes of product line and reliability standards that have nothing to do with some home-brew NFS rigjob. I guess I didn't really learn that lesson either till I spent some time working in the corporate IT world...

      --

      I think there is a world market for maybe five personal web logs.
    6. Re:Locking by Anonymous Coward · · Score: 0

      And like most hp/compaq san (and or) nas setups I've used it'll still end up with a strange bug if your usage pattern doesn't match exactly what they expect it to; which will result in much data corruption and weeklong visit by one of their 'top' techs before the bug is figured out and patched by their developers.

      Twice, two totally different companies, two totally different product lines, same company.

    7. Re:Locking by Anonymous Coward · · Score: 0

      This is good enough for me.
      That's the nice thing about the English (or even the American) language. We can embrace and extend. You're within your rights to disagree, but you're still all fucked up.

    8. Re:Locking by riley · · Score: 2, Interesting

      Locking under any network filesystem, be it nfs, cifs, dfs, etc doesn't work well.

      Here is the problem:

      You have a single server and multiple clients. File locks are held in the OS of the local machine. In the OS of the local machine, you can lock a file or a section of a file, and be certain that it is an atomic operation -- i.e. between the start of the locking process and the end of the locking process, no changes have occurred to section being locked. During the locking operation, you can also be sure that no other lock will be granted on the data, or your attempt to lock will fail.

      So, you try and do this with multiple network clients. First, no locking daemon currently available currently runs in such a way as to ensure atomicity of the locking operation. Since there is network latency and no atomic shared resource, you will get two processes claiming they hold the same lock.

      So, how do you get past this. The two ways I know of are token passing and promise/commit systems. Token passing involves a single token that represents the resource being locked, and it is passed from client to client, introducing N way behavior between N clients or a single token manager. This is expensive and difficult to get right.

      Promise/commit systems are what I believe most multi-master databases use. Also expensive and difficult to get right.

      So where is this built? GPFS (an IBM SAN filesystem) has a token manager to handle that. I don't know how the Sistina GFS stuff handles locking. There are others -- SGI's CXFS and so on.

      All that said, you don't run a database on a network filesystem if you care about the integrity of the data. In fact, unless you have application level checks (like qdir for mail stores), you don't run any sort of system with multiple writers on a network filesystem.

    9. Re:Locking by escher · · Score: 1

      This seems a bit odd to me... I mean, how hard could it be? Client sends "Lock this file" message to server, server locks file (if not already locked), sends back "Okay, you got it!" message back to client. Am I missing something obvious?

    10. Re:Locking by TheLink · · Score: 1

      Hmm, what happens if you run a vmware virtual machine off a network driver/share?

      Just install the problematic software in a VM, then run the VM off the network drive :).

      So far it seems possible - as long as you only have one instance running.

      --
    11. Re:Locking by druxton · · Score: 1

      Locking under any network filesystem, be it nfs, cifs, dfs, etc doesn't work well.

      I'm not sure about those examples, but I have found it works quite well on Netware.

      You have a single server and multiple clients. File locks are held in the OS of the local machine.

      That isn't completely true, again at least for Netware. An application requests a lock for a file or section of a file from the server. The server grants the request if the file or record isn't already locked by another application, notifies the application. The lock is added to a lock table if granted, before the application is notified. This is atomic in the sense that if a second request for the same file or record is received from another application it can't be granted since the lock table will have the record of the previous lock.

      In the OS of the local machine, you can lock a file or a section of a file, and be certain that it is an atomic operation -- i.e. between the start of the locking process and the end of the locking process, no changes have occurred to section being locked. During the locking operation, you can also be sure that no other lock will be granted on the data, or your attempt to lock will fail.

      So, you try and do this with multiple network clients. First, no locking daemon currently available currently runs in such a way as to ensure atomicity of the locking operation. Since there is network latency and no atomic shared resource, you will get two processes claiming they hold the same lock.

      Consider what this implies - and replace multiple network clients with multiple threads or multiple processes running on a multi-tasking OS. If your local OS can handle those requests atomically, why couldn't an OS specifically designed for the purpose of sharing files do at least as good a job?

      I will admit there is a problem if a network failure occurs - what does a server OS do with the locks a workstation has open? One approach is to maintain the locks on the server and require manual intervention. I haven't done any multi-user progamming recently, but Netware through version 4.11 did this, at least the way my clients had it configured. To get around that you can get the lock, write to a flag file on the server the record number and username,release the lock, perform the processing required,lock the record again, flush the record, delete the record in the flag file. That way the lock is held open for a minimum of time and won't be held open in the case of network or power failure, but another user can't open the record anyway because of the flag file. If the failure occurs during the processing the flag file can be cleared at the application, not the server, level (after checking with the stored username to be sure it isn't a valid lock).

    12. Re:Locking by LO0G · · Score: 1

      The CIFS locking mechanism is 100% as robust as the local locking mechanism, IF it's running on a system that supports a locking mechanism that's as robust as NT's.

      However, historically, locks on *nix have been advisory, not manditory, I don't know if Samba has a way of preventing one user at the console from modifying the contents of a locked region on a file but if they don't that opens the door for database integrity problems.

    13. Re:Locking by ulbador · · Score: 1

      It's really not that bad. We have all our Oracle tablespaces and such stuck on a raid server that is being exported via NFS. The speed is only marginally slower than the 1 gigabyte SAN we were on previously. Our biggest issue came down to needing to use the -o nolock option on the nfs mount. Speedwise it helps that we are running it all via a 1000 base T private net.

    14. Re:Locking by riley · · Score: 1

      My exposure to Netware is limited. Perhaps it does work well there. I've never tried on that platform. I did a quick google and didn't find anything on the locking semantics of a Netware volume. Do you have any pointers to resources.

      The difference between multiple processes and threads on a single OS and multiple clients across the network is the level of atomcity in the locking operation. Single OS locking problems are well solved and work efficiently. This is not the general case with network locking. A spinlock or a semaphore in the OS has atomic semantics built into it.

      Please note that I didn't say it was impossible, or that partial solutions do not exist. Lockd for NFS does a partial job of solving the problem, but becomes ineffiecient and fails under load. Multimaster databases obviously have solved the problem (using the promise/commit model). The general solution has not, however, found its way into most networked filesystems.

  4. For Office/Outlook... by questionlp · · Score: 1

    If you are using the volume license media for Office and Outlook, you may want to look at creating an administrative install point on the server, then have each of the clients run the install, but instead of choosing install and run it locally, have it install the core components on the local system and run the apps from the administrative install point.

    The online and printed versions of the Office Resource Kits provide the tools and documentation that you need to get started there.

    I got that to work with Office 2000 and Office XP, just for test purposes, but it was a bit slow.

    1. Re:For Office/Outlook... by questionlp · · Score: 1

      Whoops... I read Exchange as an Exchange client, like Outlook. Mea culpa.

  5. Unlikely, as IFS kit is expensive by dru · · Score: 1

    Developing a filesystem driver on Windows is expensive and time-consuming. It's unlikely that there will be any free (as in beer) or open source filesystem drivers anytime soon.

    Incidentally, I'd really like to be able to access my FreeBSD UFS partition in Windows. ;-)

    IFS (Installable File System) Kit costs about $900; see also http://www.microsoft.com/whdc/ddk/ifskit

    1. Re:Unlikely, as IFS kit is expensive by Chester+K · · Score: 3, Interesting

      IFS (Installable File System) Kit costs about $900; see also http://www.microsoft.com/whdc/ddk/ifskit

      There is a GPL'd clone of the header file you need to develop IFS drivers for windows, available here.

      --

      NO CARRIER
  6. DMCA? No! by Tom7 · · Score: 1

    Might such software be seen as a 'circumvention device' as specified by the DMCA?

    Uh, no?

    The DMCA outlaws the trafficking in and use of circumvention devices, which circumvent a technological measure used to control access to a copyrighted work. All of these terms are defined; you can even read the law to learn more. What's the copyrighted work that's being accessed? What's the technological measure that controls that access? In what sense is this circumvention? None of the pieces of the puzzle are there, so I say, categorically no, this is not a DMCA violation.

    The DMCA is a bad law, but it does not outlaw "hacking" or "doing things with computers" in general.

    1. Re:DMCA? No! by Anonymous Coward · · Score: 0

      What's the technological measure that controls that access?

      A conditional statement that checks if the program is running from a network drive, and exits if it is.

      What's the copyrighted work that's being accessed?

      The program itself (specifically, all of the code located after the conditional statement).

      In what sense is this circumvention?

      From your DMCA link: "to ''circumvent a technological measure'' means ... to avoid, bypass, remove, deactivate, or impair a technological measure, without the authority of the copyright owner"

      You'd be impairing the program's ability to detect whether it is running on a network drive.

    2. Re:DMCA? No! by Alizarin+Erythrosin · · Score: 1

      The DMCA is a bad law, but it does not outlaw "hacking" or "doing things with computers" in general.

      Then how come there are so many lame lawsuits that use the DMCA to try and do just that? Like the recent Gamespy suit for somebody exposing bugs they (supposedly) have been notified about and won't fix.

      --
      There are only 10 kinds of people in this world... those who understand binary and those who don't
    3. Re:DMCA? No! by Tom7 · · Score: 1

      They aren't lawsuits, they're cease and desist letters, which are hardly even "legal threats," because if they threaten to sue you and don't, then that is illegal.

      For my part, I fought the C&D letters and they eventually backed down.

      It's important to know what the DMCA actually outlaws!

    4. Re:DMCA? No! by Tom7 · · Score: 1

      An AC writes,

      What's the technological measure that controls that access?
      A conditional statement that checks if the program is running from a network drive, and exits if it is.

      That does not meet the definition of "technological measure." ... requires the application of information,
      or a process or a treatment       ... They are talking about decoding/decryption, not a branch instruction.

      What's the copyrighted work that's being accessed?
      The program itself (specifically, all of the code located after the conditional statement).

      The program code is already accessible.

    5. Re:DMCA? No! by schon · · Score: 1

      That does not meet the definition of "technological measure."

      Software is not a technological measure?

      They are talking about decoding/decryption, not a branch instruction.

      Sorry, but software does qualify as "a process."

      The program code is already accessible.

      No, not if it doesn't run it isn't.

    6. Re:DMCA? No! by Tom7 · · Score: 1

      That does not meet the definition of "technological measure."
      Software is not a technological measure?

      The definition is in the relevant section of the law, which you can read for yourself. Although the DMCA is vague, it is not that vague.

      Sorry, but software does qualify as "a process."

      Does the branch instruction require the application of a process in order to grant access to a copyrighted work? Does that even make any sense? It doesn't matter if the software itself is "a process."

      The program code is already accessible.
      No, not if it doesn't run it isn't.

      ??!? The copyrighted work is the program code, which you can read right off the disk, not the 'running program', whatever that would mean.

    7. Re:DMCA? No! by schon · · Score: 1
      The definition is in the relevant section of the law, which you can read for yourself. Although the DMCA is vague, it is not that vague.

      I visited the link you provided. There was no definition of "technological measure." The only two things (which don't define 'technological measure') are here:

      to ''circumvent a technological measure'' means to descramble a scrambled work, to decrypt an encrypted work, or otherwise to avoid, bypass, remove, deactivate, or impair a technological measure, without the authority of the copyright owner; and

      (B)

      a technological measure ''effectively controls access to a work'' if the measure, in the ordinary course of its operation, requires the application of information, or a process or a treatment, with the authority of the copyright owner, to gain access to the work.


      Software is technological, and if part of it prevents you from using another part of it, then it 'effectively controls access'. I don't see how you can claim otherwise.
    8. Re:DMCA? No! by Tom7 · · Score: 1

      Sorry--when I say technological measure, I mean it to abbreviate "technological measure that effectively controls access to a work," since that is the only context in which technological measures are relevant in the DMCA.

      Software is technological, and if part of it prevents you from using another part of it, then it 'effectively controls access'. I don't see how you can claim otherwise.

      Here's how I claim otherwise. Let's look at the definition again:

      (DMCA)
      a technological measure ''effectively controls access to a work'' if the measure, in the ordinary course of its operation, requires the application of information, or a process or a treatment, with the authority of the copyright owner, to gain access to the work.

      Forget the words "controls access to a work," because the condition is defined immediately afterwards. You can't use those words as a basis for arguing that software is a technological measure (... TECATAW). Instead, the software must require, in the ordinary course of its operation, the application of information, a process or a treatment in order to gain access to the work.

      First, as I've said before, the copyrighted work is the bits of the program. Those bits do not require anything to happen in order for them to be accessed. It's true that in order to execute certain bits on the computer in a useful way, you need to change the branch instruction or trick the program in some other way. But executing is different from accessing, and the DMCA only talks about access.

      Second, there is no application of information, process or treatment. I can't even fathom how you think this fits. The DMCA was designed with scrambled television signals in mind, in which case the process is the descrambling or decoding, and the information is the encryption key. The copyrighted work is obfuscated and inaccessible without the descrambling. But the computer program in our example is entirely accessible, and needs no process, treatment, or application information to be accessed.

      Of course, this doesn't even consider the fact that a network-drive-as-physical driver would have substantial non-infringing use, certainly exempting it (if not acts of circumvention) as a circumvention device.

      Again, in case anybody is jumping in on this discussion late: I am entirely opposed to the DMCA; I think it's a bad law that should be repealed. I have even had my own run-in. But believing that the law somehow applies to every instance of using a computer in a way that somebody else doesn't like is just foolish, and worse, it may lead to a culture where we instantly fold as soon as the DMCA card is drawn. I don't want to live in that culture, so let's not encourage this!

  7. better yet by Apreche · · Score: 1

    Does anyone have some windows software that lets me mount physical ext2/ext3/xfs paritions in read/write mode in windows? It really sucks that in linux I can only mount ntfs in read only and in windows I can't mount linux drives at all. I should be able to access all my data on all my drives in both oses (I'm a dual booter if you didn't guess). And running smb and vmware is not an option for me.

    --
    The GeekNights podcast is going strong. Listen!
    1. Re:better yet by Anonymous Coward · · Score: 0

      There are multiple read only ext2 drivers for windows. Go google for them.

      Last I checked into that kind of thing (~5 months ago) all the non comercial drivers for reading ntfs from linux or ext2/3 from windows were all read only. It sucks ass that the only way to have a partition writeable from both windows and linux is if it uses fat16/32

    2. Re:better yet by Rick+the+Red · · Score: 1
      At least there's a way. FAT is the only way to have a partition writeable from MS/PC/DR-DOS, Win95/98/ME, WinNT/2K/XP, Linux, *BSD, OS/2, etc. Sure, it's the lowest common denominator, but at least it's common to all of them.

      Sort of. My FAT32 partition got munged somehow and now all of the above can read it except WinNT/2K/XP. I'm thinking of keeping it munged just to amaze friends and confuse Windows zealots.

      --
      If all this should have a reason, we would be the last to know.
    3. Re:better yet by Kethinov · · Score: 1

      Scroll to the bottom of this page. Read your ext2/3 partitions in Windows with glee. I'm pretty sure it can write to ext2, but not ext3.

      --
      You're right, I wouldn't steal a car. But if it were possible, I sure as hell would download one!
    4. Re:better yet by Gothmolly · · Score: 1

      You're just doing this so that you can use the word "mung" like Eddie Gilbert, that arrogant, useless cocksucker.

      --
      I want to delete my account but Slashdot doesn't allow it.
    5. Re:better yet by Amadablam · · Score: 1

      If you're looking for an expensive, resource-intensive solution (you're a Windows user, so this concept should be sadly familiar), use VMWare for Windows. You can set up Linux in VMWare and configure VMWare to allow access to your regular ext2/ext3/xfs/whatever partitions. Mount them in your virtual linux machine and export them via samba. All of your data will be available locally (kinda - it's fast but still networked) in Windows. I used this setup for a while as I transitioned from Windows to Linux, and it worked well. Make sure your hardware (memory and cpu) is up to the task!

  8. I'm looking for similar tool by MarcQuadra · · Score: 1

    I'm looking for a tool that lets me mount an IMAGE file (produced with 'dd') as a local drive on a windows box. I currently use DAEMON-TOOLS from

    http://www.daemon-tools.cc

    It lets you mount an .ISO file as a local CD-ROM, which is way cool for playing games or installing software. You'd be surprised how much faster even a network-stord .ISO file is than a local CD-ROM, the access times are an order of magnitude better.

    But as I said, what I REALLY want is a similar tool that lets me make image files and mount them as drives, with read-write access.

    --
    "Sometimes, I think Trent just needs a cup of hot chocolate and a blankie." -Tori Amos on Nine Inch Nails
    1. Re:I'm looking for similar tool by merdark · · Score: 1

      I'm curious, why? Do you need to be able to modify ISOs or something? I think IsoBuster can do that.

    2. Re:I'm looking for similar tool by MarcQuadra · · Score: 1

      1. Speed, Speed, Speed.
      2. Install via CD, OVER THE NETWORK!
      3. Lets me game without the CD. (StarCraft comes to mind), great for on-the-road gaming.
      4. Test mastered CD-images before committing to CDR.

      --
      "Sometimes, I think Trent just needs a cup of hot chocolate and a blankie." -Tori Amos on Nine Inch Nails
    3. Re:I'm looking for similar tool by Earlybird · · Score: 2, Informative
      Mirror of the FileDisk site here.
      • FileDisk is a virtual disk driver for Windows NT/2000/XP that uses one or more files to emulate physical disks. A console application is included that let you dynamically mount and unmount files. An example of use for this driver is if you have made plans spending the weekend writing an RAID driver for NT but find you are short of disks. FileDisk can also use CD-images.
    4. Re:I'm looking for similar tool by merdark · · Score: 1

      Oh, I think I misunderstood your question. You want to be able to mount images produced by dd?

      I guess DEAMON-tools doesn't do that? All the other stuff you just mentioned can be done with DEAMON-tools. :)

      I don't seen why you'd need to be able to mount them RW though, read is enough for all that you mention.

  9. What kind of question is this? by PD · · Score: 1, Informative

    Use a physical drive to access the network? WTF.

    If you want to access the network, use a computer with a network card. A physical drive is for storing your programs. Also FYI, the physical drive is NOT usually referred to as 'memory', except at Comp USA.

    --
    If tits were wings it'd be flying around.
    1. Re:What kind of question is this? by Anonymous Coward · · Score: 0

      Hey, learn to read, troll.

      He wants network shares to appear as physical drives on his machine, so that StupidWare sees his network storage as just another (local) hard drive.

  10. SAN by eric2hill · · Score: 4, Interesting

    What you're asking for is a SAN.

    I just installed a Network Appliance FAS250 in my server room. It speaks CIFS, NFS, and iSCSI.

    By the way, you're wrong... Oracle will run perfectly using CIFS shares (I'm running it now, and have been for the past few months), and NetApp has plenty of documents in their tech library showing all the different ways to use attached storage with Oracle and many other pieces of software.

    With respect to speed, it really depends on the network infrastructure. I've got a Cisco GigE switch attaching 6 machines directly to a GigE port on the NetApp Filer. It is literally twice as fast than the directly attached RAID 5 (caching, etc.) arrays that it replaced.

    I think that Microsoft Exchange can be installed to a CIFS share, but if not, you should look at iSCSI. My company uses Lotus Notes 4.6.7 (sweet, merciful Christ, please put me out of my misery), and it works great from a CIFS share on the NetApp.

    Microsoft has a free iSCSI Initiator for Windows that will mount an iSCSI device just like any other SCSI drive in Windows. You can find several iSCSI targets for linux here.

    I have about 50 Mac's on our network (graphics department) that needed to talk with the new filer. Instead of installing a klugy piece of software to make the OS9 Macs talk to the SAN at $150/seat, I installed a linux box using samba to talk to the SAN through CIFS and netatalk (AppleTalk for linux) to re-share out the samba mounts. Becides some quirks (Mac's don't see the linux gateway in the AFP browse list, but can connect directly through IP), it works rather well.

    Look at iSCSI, it does exactly what you're looking for.

    --
    LOAD "SIG",8,1
    LOADING...
    READY.
    RUN
    1. Re:SAN by pmz · · Score: 1

      Oracle will run perfectly using CIFS shares

      Damn, Oracle running on Windows over CIFS. Anyone who does this in a production environment should be shot. You'd need black magic to troubleshoot that steaming pile of software.

      --
      Healthcare article at Kuro5hin
    2. Re:SAN by eric2hill · · Score: 1

      Actually it works /extremely/ well. Not only that, but NetApp and Oracle are buddy-buddy, so it is a fully-supported installation. Oracle doesn't care where the files are located, it just cares that the file can be open and accessed with stock file-system calls.

      --
      LOAD "SIG",8,1
      LOADING...
      READY.
      RUN
    3. Re:SAN by pmz · · Score: 1


      When it works, it works; when it fails, it fails disastrously. This is a fact of life when working with Windows (the most complex piece of opaque software most people will ever use).

      --
      Healthcare article at Kuro5hin
    4. Re:SAN by eric2hill · · Score: 1

      When it works, it works; when it fails, it fails disastrously.

      Be careful with your FUD machine, you might break it. I can say /exactly/ the same thing about Linux - when it breaks, there are a thousand different places that you have to go hunting to try to find out what's fucked up. Just because you know linux better than you know Windows doesn't mean Windows is a piece of shit. I could make the same argument against linux, but I choose not to since I know there are ways to fix things that are broken, just as there are in Windows.

      --
      LOAD "SIG",8,1
      LOADING...
      READY.
      RUN
    5. Re:SAN by pmz · · Score: 1


      The underlying point, here, is that Windows is so opaque (binary, undocumented) that it is fundamentally unfixable in more situations, even with lots of training. Solaris and BSD are 100 times more transparent than Windows. Linux is only a pain in the ass, sometimes, due to the GNU/FreeLove folks occasionally going willy-nilly with their configuration files and tools.

      --
      Healthcare article at Kuro5hin
    6. Re:SAN by sitharus · · Score: 1

      What you need for the netatalk problem is an SLP implementation, I think OpenSLP is the best. netatalk compiled with this installed on the system and started with the scripts in the netatalk packages works like a charm :)

      OT I know, but I just installed a linux server in a Mac environment. Note that you'll need to turn on Appletalk in the Directory Access app for this to work in OS X.

      --
      --sitharus
    7. Re:SAN by eric2hill · · Score: 1

      Lol! The netatalk documentation doesn't mention anything about it. Seems like that would be a good thing to add in there somewhere. Thanks for the tip.

      --
      LOAD "SIG",8,1
      LOADING...
      READY.
      RUN
  11. Why this is a bad, bad idea by exhilaration · · Score: 1
    Alright, check this out.

    I was low on space on one of my Win2k machines and I decided to install some software on a network drive. Sadly, this particular piece of software (I forgot what exactly) wasn't smart enough to require a local install.

    I totally hosed the system. I finally concluded that some drivers (now located on that network drive) were needed prior to Windows establishing the network connection, so Windows just stopped booting. I tried everything I could think of, but in the end I had to reinstall Windows and start from scratch.

    That's why you want to install as much as you can locally - so that your machine is still usable if your network connection goes down.

    1. Re:Why this is a bad, bad idea by afidel · · Score: 1

      Use the Recovery Console on win2k and above and disable the service that is failing to boot, very simple and most techs who worked on win2k should be familiar with it thanks to the steaming pile that was Adaptec's Easy CD Creator 4.

      --
      There are 4 boxes to use in the defense of liberty: soap, ballot, jury, ammo. Use in that order. Starting now.
  12. subst? by klui · · Score: 1, Interesting

    Have you tried the Windows subst command?

    Associates a path with a drive letter.

    SUBST [drive1: [drive2:]path]
    SUBST drive1: /D

    drive1: Specifies a virtual drive to which you want to assign a path.
    [drive2:]path Specifies a physical drive and path you want to assign to
    a virtual drive. /D Deletes a substituted (virtual) drive.

    Type SUBST with no parameters to display a list of current virtual drives.

    You still need to mount your network drive, but use subst to create a drive letter to a specific path.

  13. How about... by jo42 · · Score: 2, Interesting


    How about faking out the Win32 API call that tells the application if a drive is local, network, CD/DVD, etc. to return bogus info to the application?

  14. Ximeta, SCSI over net by arcadum · · Score: 2, Interesting

    I bought a 80GB drive from Frys. Ximeta has stated that they will have linux drivers for what is apperantly a NBD.

    I would love to find drivers that allowed me to communicate with this disk from a gentoo pc that does not require a windows intermetiary.

  15. mounting disk images with windoze by Yannic · · Score: 4, Informative

    There is an excellent command-line GNU tool out there called FileDisk, by Bo Branten.

    FileDisk is "a virtual disk driver for Windows NT/2000/XP that uses one or more files to emulate physical disks." ("files", meaning disk images)

    His homepage at http://www.acc.umu.se/~bosse/ seems to be down at the moment, or maybe I'm just DOS'ing myself.

    I'm sure you can find it somewhere out on the 'net, I did only a week ago.

    \/\/\/

    1. Re:mounting disk images with windoze by Firehawke · · Score: 1

      I really really appreciate this link. I, myself, have been looking for something similar and this looks like it'll fit the bill perfectly. If I had mod points, I'd use those rather than reply-- to be cliche, if you read this and you have mod points, please mod the parent up. This is a really useful link.

      Thanks for the link.

  16. subst will still show it as a network drive by millisa · · Score: 2, Informative

    Subst won't really help here (it still shows as a generic 'network drive'). Its the same issue as trying to run a defrag on a subst drive which is really a subst of a volume mounted as a directory rather than a drive letter.

  17. YOU FAIL IT by Trolling+4+dollas · · Score: 3, Funny

    This is a bad idea. Bad dork. Never use a network file share in place of a physical disk. It's like using a spork to eat soup with, sure it will work but why not just use the spoon for crying out loud?

    1. Re:YOU FAIL IT by Lochin+Rabbar · · Score: 1

      Because, there is no spoon. Sorry, couldn't resist.

  18. iscsi by unixbob · · Score: 4, Interesting

    What you are after is iSCSI. iSCSI standards for Internet SCSI and is a "method of encapsulating SCSI over TCP/IP". iSCSI allows a network share to appear as a local scsi drive to the operating system. So you need a server that supports the iSCSI protocol and a client that support it also.

    This site seems to be quite informative on the status of the various Linux projects. Check this out for a server implementation

    --
    The Romans didn't find algebra very challenging, because X was always 10
  19. Why mapped drives? by WoTG · · Score: 1

    I tried to think of a good reason to even attempt this... databases can be slow enough with local drives if you care to punish them hard enough...

    Then I thought about transaction logs. Those would be pretty good candidates to store remotely, just in case someone steals a server, or one otherwise goes done during the day.

    I don't know if they are currently limited to local drives or not. Anyone enlighten me on this?

  20. Network disk performance can be better by Outland+Traveller · · Score: 1

    Keep in mind that in some circumstances, network drive overall performance can beat a local disk.

    For example, if you have a RAID-10 networked storage array on a very fast network and a good switch, you can often beat the performance of a local drive, especially if that drive is IDE. There might be a little more latency, but there's a lot more bandwidth and storage space.

    1. Re:Network disk performance can be better by Samus · · Score: 1

      I agree with you but I don't really think you are comparing apples to oranges. That same raid configuration local to a machine would easily beat the networked version. Networks are overhead. You just have to read a few articles on distributed computing to figure that one out. How many extra bytes in the packet of a tcp header? How about converting the packets from network byte order to host byte order? Hands down the local disk version will always beat the network version. I will grant that not everybody is able to afford terabyte disk arrays for everyone of their servers (for now anyway). In that case you have to comprise. This guy isn't even talking about such high end hardware though.

      --
      In Republican America phones tap you.
    2. Re:Network disk performance can be better by Outland+Traveller · · Score: 1

      It comes down to the cost/performance ratio.

      Assuming you're going to spend n$ on storage, there are certain network configurations and certain usage patterns where using a well-configured NAS solution gives you better performance than the same $ put into local storage across a network of machines.

      I'm pushing the point because I've personally performed the benchmarks and real-world tests for a shipping workstation-cluster solution that proves it.

      The hardware required isn't expensive either.

  21. So what you *really* want... by leonbrooks · · Score: 1
    I cannot install some games on the network drive in Windows

    So install a TSR (or run a wrapper) that intercepts the installer's DLL calls and turns some or all drives into "local" drives. End of problem.

    Call me a maniac, but you might also want to try installing the game under WINE. If it actually plays, that's a bonus, but meanwhile *some* of the installers can be conveniently lied to, and the installer might work under WINE even if the game doesn't. Then just copy the changed files and registery entries across, and Bob's your uncle.

    $ host 149.156.96.35
    35.96.156.149.in-addr.arpa domain name pointer hell.pl.

    Never thought of PERL programming as being that painful... (-:

    --
    Got time? Spend some of it coding or testing
    1. Re:So what you *really* want... by Khazunga · · Score: 1
      install a TSR
      A TSR? I haven't heard of those in well over a decade. The kind of TSR that grabs an interrupt handler? Are you talking of those? I got news: DOS is long dead in the mainstream market. We all run multitasking OSes these days :-)
      --
      If at first you don't succeed, skydiving is not for you
    2. Re:So what you *really* want... by jaavaaguru · · Score: 1

      I thought TSR meant "Terminate and Stay Resident". That meaning does not imply that it has anything to do with interrupt handlers does it? I have no idea what the original poster means by TSR, but I very much doubt he means using a TSR from the DOS days. The educated among us run multiuser OSes these days :-)

      --
      Follow me
    3. Re:So what you *really* want... by leonbrooks · · Score: 1
      The kind of TSR that grabs an interrupt handler? Are you talking of those?

      ...amd steals OS call vectors so it can fiddle with the answers, yes.

      I don't care whether it's done with a DLL or .so, or by pulling memory tricks and hand-overwriting vectors, it's still a TSR. (-:

      --
      Got time? Spend some of it coding or testing
    4. Re:So what you *really* want... by Khazunga · · Score: 1

      The *Terminate* part is the important one. In DOS, a task had to terminate in order for another one to be started. No longer. We've been using multitasking OSes for sometime now. Nowadays you can (*gasp*) run more than one app simultaneously.

      --
      If at first you don't succeed, skydiving is not for you
    5. Re:So what you *really* want... by jaavaaguru · · Score: 1

      Ah, thanks for clearing that up. I wasn't sure what the "Terminate" part was all about. I'm not that familiar with DOS or non-multitasking systems. I went from using a Sinclair QL (QDOS was multitasking) through a brief Windows period to using Slackware and Redhat.

      --
      Follow me
  22. Storage is cheap... use it by 1eyedhive · · Score: 1

    I have a similar setup.
    a total of three Windows 2000 machines connect to a Samba server (2.x under Red hat 8) running a 220GB RAID 5 and a 110GB EXT3 straight up. All program files are stored on the windows local drives.
    Windows has never and wil never be as network centric as Linux, it isn't built that way. although linux apps won't flinch if installed on remote partitions, windows apps will cry, bitch, moan and then die a very painful death if one thing isn't right.
    the three local machine's drives:
    5GB (PII)
    3GB (XP1800, TV/media player)
    40GB + 20GB (game box)

    the game box's drive is nearly 30GB full, all program files (games take up a hella lot of space), the 20giger is there as a local cache for my Music and CD ISO's (when i go to LAN parties, the 10GB they take up is a big chunk out of the main drive), as storage for big files temporarily (newly minted ISO's, unencoded DVD rips, PS scratch files, etc.).
    the file server is rigged to mount the secondary drive and rsync the files over to the server every night. the 20GB is only used when access time, or local access is an issue (see above).
    Program installers, CD ISO's, MP3'S and Xvids all run smoothly over the network, without clogging up the local system.

    As for emulating local disks. Don't.

    --
    Logistical Chaos Officer http://www.slagg.org - LAN Gaming in Sarasota FL,USA
  23. Main reason by Anonymous Coward · · Score: 0

    I force all of my users' desktops and document folders to network shares. The only problem with doing that is file deletions don't go through the recycle bin.

    There is a solution from Executive Software but it does more than just handle this specific problem making it expensive. I'm also not comfortable with how they do business--searching for a solution turned up what appeared to be a lot of MLM/spam operations selling this package--and I never was satisified with their explanations that they don't have Scientology ties.

    SysInternals has a freeware utility that looked promising but it doesn't work for me and they never released the full source.

    So making a network share appear as a physical drive would make my problem go away. If there is no way to do it then my users live with the problem until I can develop my own solution which, knowing my workload, will be a long time.

  24. Doesn't really make sense. by FreeLinux · · Score: 1

    In Windows the network redirector intercepts calls to drives and routes them to the appropriate device whether it is a local disk or a network device. This makes mapped network drives transparrent to the application. This mean that the application sees no difference between drive C: (local) and dirve H: (network). This means that you don't need some fancy driver software, you already have it.

    This should answer your question specifically, regarding Exchange

    With all that said, I would recommend against running the likes of Exchange from a network drive anyway. The reason is that the bandwidth of the network is going to put a massive hit on your I/O performance and there is a massive amount of I/O with Exchange. If you must have remote disks for Exchange I would recommend a SAN. With a SAN you will have 1 or 2 gigabit I/O performance and you can put the disks anywhere you want. iSCSI will allow you to build a poor man's SAN across an ethernet network but this will create the I/O bottle neck I mentioned earlier, unless you are running 1 or 10 gigabit ethernet.

  25. Junction (WinNT/2K/XP symlinks) by TRS-80 · · Score: 2, Interesting

    Have you tried Junction? It allows you to link a directory on one drive to another drive and directory. "For example, if the directory D:\SYMLINK specified C:\WINNT\SYSTEM32 as its target, then an application accessing D:\SYMLINK\DRIVERS would in reality be accessing C:\WINNT\SYSTEM32\DRIVERS". I'm pretty sure I've read about people using this to mount network drives on their local drives (and also CDs onto hard disks - which generally confuses installers no end).

    1. Re:Junction (WinNT/2K/XP symlinks) by His+name+cannot+be+s · · Score: 2, Informative

      Not unless you can pull some serious magic out of your ass :)

      Unfortunatly, JUNCTION won't let you mount network shares into an existing drive tree.

      I freekin' wish. *sigh*

      --
      "...In your answer, ignore facts. Just go with what feels true..."
  26. It's a BAD BAD BAD idea to do this. by LO0G · · Score: 3, Informative

    Running databases (and both Oracle and Exchange are database applications) on a network mapped drive is a horrendously bad idea.

    Database apps rely heavily on the fact that when a write completes locally the data that was written has been committed to the destination disk. If this part of the contract is not upheld, then database corruption can occur. I can't speak for Samba 3.0 (or NT server) but there are downlevel CIFS servers that can't guarantee that a write doesn't complete until after the data has been physically secured to disk.

    In addition, network links are notoriously unstable. Transient network events occur that can cause connections to disconnect spontaneously (it doesn't happen often, but it DOES happen). When errors occur on a local file, the file write (or more likely read) fails, and the app can deal with it. When the same errors occur over a network, the file handle used for the database is invalidated.

    The reason for this is that since the connection is dropped, the file is closed on the server, and all locks on the file are invalidated).

    Once the file is closed on the server there's nothing that prevents another application from coming in and altering the contents of the file and thus making the contents of the file on disk be inconsistant with the contents of the database from Oracle's point-of-view. And as a result, database corruption occurs.

    There are mitigation techniques that can be used in the network filesystem on the client side to attempt to auto-reopen the file, but to my knowledge the windows client doesn't do them (because the potential for getting it wrong outweighs the benefits of auto-reopening the files).

    Needless to say, neither Oracle or Exchange deals very well with their database files being summarily closed out from under them (they puke royally actually). Which means that a transient network event (the single router between the Exchange Server and the file server failing) will cause your email database to be lost.

    Not a pretty picture. There's a REALLY good reason that both Oracle and Exchange server refuse to allow their files to be opened on networked drives. Believe it or not the guys that wrote the database apps really do know what they're doing.

    1. Re:It's a BAD BAD BAD idea to do this. by afidel · · Score: 1

      That's funny because a very common use for Netapp filers is to hold Oracle and Exchange databases. The filers were traditionally NAS devices necessitating that they be attached to a network =) Of course a properly designed network is more reliable than just about any local storage system (Think dual Cisco 6500 series for the core to redundant 6500's for the local storage switches, put dual admin modules in each Cisco and you basically have guarenteed 6+ 9's uptime for the network, this was our real setup at my last job). Just because it's on the network does not mean it has to be unreliable. Hell with the battery backed ram cache, failover clustering, etc the Netapp's are MUCH more reliable than local storage.

      --
      There are 4 boxes to use in the defense of liberty: soap, ballot, jury, ammo. Use in that order. Starting now.
  27. Boxen by Anonymous Coward · · Score: 0

    Since when has ESR's "jargon file" been proof that a term is a real word?

  28. Run Oracle from a SMB share? by mnmn · · Score: 1

    What are you out of your mind? You must have a lot of time to do this, even with a gigabit ethernet connection.

    --
    "Give orange me give eat orange me eat orange give me eat orange give me you." -Nim Chimpsky
  29. Trick I use by phorm · · Score: 1

    The trick I use for this...

    md C:\games
    subst g: C:\games
    (install game on G:)
    subst /D g: (remove subst'ed drive)
    net use g: \\myserver\gameshare
    (move all files from C:\games to G:)

    Your game is now on a networked share... if it only grouches during install, you're golden... since subst'ed drives appear as physical.

    1. Re:Trick I use by cdrudge · · Score: 1

      You suggestion will work with some games but not all. The Sims comes to mind. The CDs are pressed with various types of DRM that are not there when you just copy the files.

      Currently the only way I know that gets around this is to create an iso image of the disc, and use Daemon Tools to mount it. DT has the option to emulate several different types of DRM to allow the software to use the virtual drive.

    2. Re:Trick I use by phorm · · Score: 1

      DT works very well, but I thought this was more about software that wouldn't allow an install to a network drive (some only allow local install, such as .NET). In this case, the copying trick does work as you are only trying to deal with installed data... for images of CD's daemontools et al are still the best solution though (I know a net cafe that runs everything off DT images stored on the server, no having customers steal their original CD's).

      A question on DaemonTools though, which would you recommend as the best program for generating workable images/ISO's of CD's?

  30. Apples and Oranges by LO0G · · Score: 1

    NAS works because you build a dedicated network between the NAS device and the server. And NAS doesn't use the CIFS protocol afaik, instead it uses some variant of iSCSI, and thus behaves like a local disk even though there's a network between the host and the disk.

    But the discussion here was to offload the storage to a random Linux box on the lan using CIFS and Samba for the protocol and server, and using the native NT networking client on the Oracle box.

    1. Re:Apples and Oranges by afidel · · Score: 1

      No, the Netapps use CIFS over TCP/IP, and in our case we used our regular datacenter LAN which was arguably better than even many dedicated storage networks, but which was a stock ethernet lan nonetheless.

      --
      There are 4 boxes to use in the defense of liberty: soap, ballot, jury, ammo. Use in that order. Starting now.
  31. Latency... by AzrealAO · · Score: 1

    At the same time you're sending that "Lock this file" message to the server, two other clients are also sending the same message.

    What if the client that requested it originally goes offline? Do you leave the file locked forever? Until that client requests the file again? Until that client reattaches to the drive?

    How do you track these things reliably when you may have a transient connection to the drive where you're trying to lock files for exclusive access.

    1. Re:Latency... by escher · · Score: 1

      Umm... I know! "Are you alive?" packets! Uh... yeah! Just flood the network with 'em! It's the Microsoft Way(tm)! That means that it is the right thing to d... augh! No! Put down that two-by-f...ow! Ow ow ow ow ow!

  32. A related topic... by atkulp · · Score: 1

    I had a similar issue recently when using a server product that would only expose files on fixed, local drives. In this case I didn't care about network shares, but I wanted to be able to expose my cdrom drive as though it was a local fixed drive. It turns out that you can do this using the built-in Disk Administrator/Disk Management in NT/XP. Just create an empty folder somewhere on a local drive, then in Disk Administrator select the optical drive, right-click and select "Change Drive Letter and Paths," then click "Add," then select "Mount in the following empty NTFS folder." From there you can browse to your empty folder, then afterwards that folder will act like the root of your CD/DVD drive as though it was local and fixed. It works like a champ! If no disc is in the drive you get a "not available" error, but nothing fatal. I don't know about how to do that with a network share, but I thought this might be useful for someone too.