Windows

Use mount points if you run out of Windows drive letters

If you're running out of drive letters, one trick is to use a mount point for each logical drive that you are going to bring into Windows; this way, performance can be contained to a logical drive and still conform to your drive letter standards.

In order to get granular control of drives and their performance within Windows, admins usually separate Windows drive letter assignments by role for better performance. However, you can only do so much with the drive letters E: and higher.

If you're running out of drive letters, one trick is to use a mount point for each logical drive that you are going to bring into Windows; this way, performance can be contained to a logical drive and still conform to your drive letter standards.

There are many scenarios in which you would want a large number of drives, such as multiple databases for Microsoft SQL Server or Exchange Server installations. Exchange databases are notorious for needing their own drives per mailbox store and, if you provision out well, you will quickly run out of drive letters. This can enable large numbers of drives to be available within one system.

Adding mount points in Windows Server 2008

With the disk available to the Windows Server 2008 operating system, right-click and select New Simple Volume. Specify the size as you would normally add a drive. Select Do Not Assign A Drive Letter Or Drive Path (Figure A). Figure A

Figure A

Perform the drive format and assign a label as normal. Once the New Simple Volume Wizard is completed, the drive will be inventoried in the list of disks. MountPoint3 has been added and now select Change Drive Letter And Paths (Figure B). Figure B

Figure B

When the Change Drive Letter And Paths option appears, point the drive to a path on the local filesystem. In this example, it is pointed to C:\MountPoints\MountPoint3. With this configuration, the path will use a different controller than the C:\ drive for the I/O operations across all drives. Figure C shows redirecting the drive to the MountPoint path. Figure C

Figure C

Once your configuration is complete, all of the mount points will appear in the subfolder you created. The "folder" icon for the mount points is a shortcut to a drive; the "folders" are redirecting the drive to the path on the parent drive. Figure D shows three mount points in one folder pulling in three different drives. Figure D

Figure D

Consolidating the drives, especially in shared storage configurations, can greatly simplify searching and organization for large systems on a Windows server. This functionality is not new to Windows, but it looks a little different now in Windows Server 2008.

Stay on top of the latest Windows Server 2003 and Windows Server 2008 tips and tricks with our free Windows Server newsletter, delivered each Wednesday. Automatically sign up today!

About

Rick Vanover is a software strategy specialist for Veeam Software, based in Columbus, Ohio. Rick has years of IT experience and focuses on virtualization, Windows-based server administration, and system hardware.

44 comments
zackers
zackers

My experience with mount points under XP is that many apps do not fully understand them. For example, the Roboform Portable installation cannot find an USB flash drive unless it has a drive letter. Another situation is the Erase program from Sourceforge which will follow the directory tree and erase the empty space on a USB drive mounted under a HDD-based drive (this very quickly shortens the life of the USB drive).

CharlieSpencer
CharlieSpencer

Sorry, I'm missing the point. Why not just create a shortcut to the share\drive\folder you're interested in accessing?

jeff.ohlhausen
jeff.ohlhausen

Hi, Anyone using this with Microsoft Clustering? Specifically in conjunction with SQL Server 2008? Thanks Jeff

bulk
bulk

I use mount points with Media Center. Normally TV recordings are stored in the user's profile in "Recorded TV". The files get quite big and numerous, so many buy a large, slowm, cool and cheap second disk. Simply set a mount point in Recorded TV that points to the new disk. One tripping point is that a subdir has to be empty before a mount point can be created. The Media Center eshell process creates files in the directory when it starts. To overcome this, restart XP/Vista MC in safe mode so that the eshell process doesn't start. Delete the files, create the mount point, restart, done! RS

The 'G-Man.'
The 'G-Man.'

Nothing New Here. This was even part of the upgrading to XP MCSE course.

kgouldsk
kgouldsk

I use mount points to allow me to build extremely consistently structured servers, regardless of the size. My OS is on C - NOTHING else, unless it can't be redirected. Protect the OS at all costs. If you fill up the disk with a temp file and it goes down, you're busy recovering the OS when you should be working on your application. D: is the full applications and data drive on small servers. Z: is the CD X: is swap The files for programs belong under D:\Apps or D:\Data. When I have a server that needs to scale up, the data still goes under D:\Data, but for any directory that is going to have a large amount of data, or has a different performance characteristic, such as database logs, I will use a mount point. This lets me maintain the core directory structure, but add capacity as applications need it. I don't have to look through my environment of hundreds of servers and remember which server has user shares on F: or Q:, group shares on G: or N:. It's always the same. Exchange or SQL will have multiple DB volumes, and multiple log volumes, which may be RAID10, RAID5, whatever, but still in the same directory structure. There are several limitations. One is the size reporting that others have mentioned - when I do properties on a mounted volume, I don't see any usage data - MS could fix this very easily, but they don't prioritize it, and I think it's unforgiveable. The second is that the recycle bin is not supported, but is still called, so when you try to use Explorer to delete a file, it hangs for a while, then errors out. The solution is to hold shift while deleting to bypass the recycle bin. I too have occasionally noticed problems deleting from the command line, but that seems to be a recent introduction of a problem, as it never used to happen. Finally, my described method above is a bit limiting, because when you add new capacity, you're constrained to adding subdirs to it to support application, so they appear under a directory that represents the disk. Recently, I've been following another method, which is to create a D:\Mounts, then mount my volumes in directories thereunder. I then use filesystem links to link each application directory back to a subdir in the mounted volume. So instead of having: D:\Data\Shared (mounted vol) \Common \Vol1 (mounted vol) \Users \Group \Vol2 (mounted vol) \Projects I now have: D:\Mounts \Vol1 \common \Vol2 \Users \Vol3 \Group \Projects D:\Data\Shared \Common ->D:\Mounts\Vol1\common \Users ->D:\Mounts\Vol2\Users \Group ->D:\Mounts\Vol3\Group \Projects->D:\Mounts\Vol3\Projects To restructure, I can mount a new volume, copy the data to a new subdir on it, and update the link. Links are a bit of a b*stard child still on prior to Server 2008, so beware of that. I've had pretty good success. At first I thought I could use the ability of the volume to be mounted on multiple directories but then realized that would mean that the 4 directories contents would all be merged together, so that doesn't work. I don't know if they've fixed the volume size reporting or delete issue, anxious to check it out - please post your results. Here's a way to report usage: C:\WINDOWS>wmic volume get caption,capacity,freespace Capacity Caption FreeSpace 20973137920 C:\ 13854384128 88266911744 D:\ 27913060352 36410552320 E:\ 7188156416 1795701968896 D:\Data\Shared\D01\ 1137574416384 Z:\

dan.nutley
dan.nutley

FINALLY a clear explanation on how to use these features to get away from those horrible "Drive Letters"! Why would I want this? Because if I need 15 different drives for Exchange, I can name them Exch1 through Exch15, and then if I need another 20 for SQL Server I can name them sql1 through sql20. And I never have to ask "let's see--is "K" an Exchange drive, or is it a SQL Server drive, or something else?" The reason you want this is the same reason you don't want to be forced to name your Word documents "A," "B," "C," etc. You want to name them "Resume," "A Portrait of the Artist as a Young Man," "Why Drive Letters Suck," etc.

axel.schmidt
axel.schmidt

Yes, this is a nice method to handle drives. One big disadvantage: you have no way to see, how many memory is in the path, you only see the available memory of the top drive

ke_xtian
ke_xtian

The sooner Windoze can get away from it's checkered past, the better. I guess you can take Windows out of DOS, but you can't take DOS out of Windows.

regular guy
regular guy

Please rewrite this with some better reasons to use drive points and more explicit instructions.

donaldvc
donaldvc

Yes, this should be no problem. We use mount points as a standard for all our SQL Server clustered installations (2005 & 2008) without any issues. One minor caution is that the Recycle Bin for the mounted volumes will be located on the root drive - not really a problem unless some of the following conditions apply: 1. You attempt to uninstall a SQL instance 2. The combined size of the system databases is greater than the free space available on the root drive. In which case, the uninstall will fail as the system databases are deleted to the Recycle Bin with an insufficient free space error. So, be careful in the sizing of your root drive!

b4real
b4real

Dan: Thank you for the sanity check. I thought the material was pretty clear, glad you get it!

it_admin
it_admin

When you go to the mount point and select properties, you get a window with several tabs. On the first tab there you can lock up the abaible diskspace for that particular drive. It works at least with WinXP Pro

mitzampt
mitzampt

Well mount points are like linux FUSE mounts, so it's good that a server product is just implementing an idea that was here since, let's say... ever! These are the downsides of backward compatibility, you end up fighting with an idea that was acceptable then but now it's getting in the way: 1.Why do you need to assign a drive letter for everything, linux uses the simple approach of grouping stuff in trees so that a mailbox path would start like /mail/mail1 I'm not a linux over windows guy, but the only letters i need are for local (physical) drives i have no name for.... and maybe for operating system root since internally you don't like to change all the stuff at os install right? 2.No windows folder except a drive with a letter will ever say you how much space is avaible. Now that should be fixed not only because of mount points, there are several gains in that starting with extending quotas from disk to shared folders and other... 3.This is a microsoft software and i know nobody will listen to what i have to say, because of bad publicity :D

Neon Samurai
Neon Samurai

- simplified directory paths without the letter mount point so your .cmd or other pointers rather than having a different drive litter assigned each time the Flashdrive conflicts with another letter mapping. - mount storage space directly under a user's directory structure. This is handy for having a shared folder apear within each user's My Documents instead of navigating to a different location. The built in Shared Documents setup may not be apropriate in some cases. - access your storage without seeing 24 letters apear in My Computer. I have a reader that support various flashdrive formats, a few firewire ports and a few USB ports. The drive slots all get assigned letters even when empty so I've always a minimum long list of drives when I open the explorer or disk manager. - mounting across a network to a local directory can present a more unified structure to the user. CIFS bound to a drive letter works but it may be even better to bind that share to a directory name within the user's space. I always build with a C: for program installs and a D: for My Documents and other data needed after a system reinstall. I could map some of those directories into my user home directory instead of always going to D: for them though. I could map my saved game folders to directories under the C: game installs saving the latest version if I have to wipe C: for some reason. Having gotten very used to mounting under a directory name, I'd have been using it long ago if introduced to the concept earlier than starting with Unix like systems. Novell did it I believe but it was already fading by that time.

davele
davele

Re Point 3. Actually Microsoft is very proactive in listening. There is a site http://connect.microsoft.com/ where you can connect. Post your feedback, Provide suggestions on what you'd like, participate in beta releases etc. It was their passion for listening to my suggestions that eventually caused me to join the company. On Mount Points specifically. They have been in Windows for many years. They were also part of various Unix implimentations prior to that. And were in available in various Mini & Mainframe O/S's prior to that. So you may ask why weren't they in the first verion of DOS? Because the H/W it ran on was very memory constrained & because abstraction creates complexity. While handy in todays equipement where multi-perhiperals are commonplace. They aren't something your average end-user wants to deal with. Evidence of this can be seen by this thread. Where many are wondering why they should be interested in a feature that has been in the OS for at least 8 years. Others the Data Centre folks "saw the need" immediately. Hope this helps & thanks for your suggestions. Dave

kmcniff
kmcniff

The only reason to require drive letters is the design of the software that is installed and running on the machine. Also drive letters are important for users as this is familiar to them and they don't have to know where that letter is connected to. If software used the same device conventions as Unix there would be no problem. As a matter of fact Windows can boot from the device path as long as the core files are on C: or the first device on the first boot controller, and the space requirements are small. From there I can direct the OS to boot from anywhere I like. This ios more of a hardware issue (Intel) than a OS issue.

Neon Samurai
Neon Samurai

I just took a quick look - the properties of a folder will give you it's size including subdirectories though it won't give you available space. - the properties of a drive will give you space used and space free. This includes space available in storage mounted across a network (CIFS). - I can't confirm if a drive includes the space of storage mounted as a subdirectory under it. That is a bit of a problem MS needs to correct. - I also can't confirm if properties on the directory storage is mounted too shows space available. Is there a cli management tool that lists mount points and used/free space? In the *nix world, one just types "df" and there it is.

O & G IT Guy
O & G IT Guy

That's funny, I am using this same feature in Windows 2003. It is not new in 2008, the author just used 2008 as the example of how to do it (although it looks pretty much identical in 2003, and I don't have anything older to verify, but I believe it was there in 2000 as well)...

TonytheTiger
TonytheTiger

when users started requesting memory card readers. We already had network drives in the G: and up range, but ran into problems when the card reader wanted F:, G:, H:, and I: We solved it similarly by mounting each slot in the card reader to an empty directory on the C: drive in Disk Manager.

kmcniff
kmcniff

I work in data centers that have clusters and sql servers that use SAN for disk space. Often times we will run into a situation that we need to move the data to another SAN frame and the Windows already has 12 or more drive letters in use. To migrate the data and move it requires a second SAN path to the new frame. If you do the math it takes 2 drives for each volume to mount disks on each SAN frame. As there are 26 letters available and A,B,C, and another for the CD that leaves 22. That means that if you have more than 11 drives then you will not be able to move the data from one frame to the other all at one time. So mount points come in VERY handy there as a necessity to maximize the movement of data and be able to copy the data all at the same time from all old frame to new frame.

Dumphrey
Dumphrey

a hard drive split into two partitions, one gets mounted as Documents and Settings (hence all users My Documents) and the other as Program Files. then I just have a small image of my os to restore if need be. And by putting this disk on a separate controller from the OS disk, some programs perform a little faster. Another good use is to mount a disk to a folder on your Desktop to allow drag and drop "archiving".

TonytheTiger
TonytheTiger

a thumb drive mounted as the "My Documents" directory. Each user has his own... You can back it up to a server as part of the logoff. Instant quota.

kmcniff
kmcniff

Treesize will tell all.

b4real
b4real

I agree, Windows hould be able to do it better than 3rd party tools. I use NTP quota and file Sentinel software, it rocks.

Neon Samurai
Neon Samurai

I'm not defending the design decisions and choice not to fix the shortcomings after the product started shipping. For space limitation, I guess I just assumed there was a clean way to do it through AD. MS is not really built to make proper use of drive partitioning though. It holds close to the legacy need for everything to work back to Dos programming. Either way, I'm not in Windows enough each day for it to matter really. Work has an IT department that won't give me the admin password so there I use it as is.

mitzampt
mitzampt

As I said, this is not only because of mount points. Say wouldn't be a terrific feature for a server to limit the size of certain folders just by assigning a limit of space... and by that i mean not like the ugly feature of disk quotas in windows which prompts you at logoff that you have to delete some of your files in order to shutdown... How many of you have gave up this feature because of that? Wouldn't it be great that, let's say, when you share a folder and someone has write access to it he wouldn't have the opportunity to fill up a 10Gb partition with junk just because after that 1gb of 'avaible space' the share will tell him 'the disk is full'? I am talking about the potential of a feature, that comes in handy. And why should we go to the root drive to see the avaible space everytime we need to copy something in a folder? This property can be inherited just like all the other properties in NTFS filesystem.

Neon Samurai
Neon Samurai

I remember hearing that many of MS own programs rely on letter mappings along with most third party programs. As a regular user, you can't open the management console and disk manager; booo! That would work for admin but it won't show a regular user how much space is hidden in that child mount point.

Dumphrey
Dumphrey

show you all of this info as will any similar utility. Most of the MS utilities only work on drives with letters... lame really.

DanLM
DanLM

;o) Have a good one. Dan

kmcniff
kmcniff

Usually because the total path to the files are too long. Use the SUBST command to assign a drive letter to a PATH and this will reduce the total characters in the full path/dir/dir/.../filename.foo to something smaller and allow the delete. I have run into this when users save emails to the personal drive and the message name is 246 characters long. This will reduce the total character \\server\path\...\filemame and the file can be renamed or deleted. Mountpoints can be used for the same thing as the SUBST command.

Photogenic Memory
Photogenic Memory

http://en.wikipedia.org/wiki/Royal_Canadian_Mounted_Police You see; back in the old days mount points were like check points( the American equivalent ) in the wilderness. In addition when bad guys needed to be caught; then all that was needed was to setup a mount point ( a drag-net ) and fan out on foot and horse back. Eventually the hooligans would be found and properly.....whatever, eh. Good night, LOL! ( just horsing around here folks. )

DanLM
DanLM

And you quite obviously missed that unix/linux had it before windows 2000. Who here is catching up to who? http://www.sbras.ru/cgi-bin/www/unix_help/unix-man?mount+2 [i]Mount() and unmount() function calls appeared in Version 6 AT&T UNIX.[/i] http://kb.iu.edu/data/agjt.html [i]Versions 6 and 7 (aka V6 and V7) represent the foundations upon which both major Unix flavors (BSD and System V) and all modern implementations are based. Their names refer to editions of internal documentation at Bell Laboratories, and the two releases are sometimes referred to as the Sixth and Seventh Editions. Neither V6 nor V7 were commercial products, nor were they officially supported by Bell Laboratories. [b]Version 6 (1975)[/b] was the first Unix distribution widely available outside of AT&T; academic and research institutions could obtain the C source free of charge. In particular, the University of California at Berkeley made extensive modifications and improvements to the code, which became the Berkeley Software Distribution (BSD). In 1979, Bell released Version 7, which incorporated some of Berkeley's improvements and was the first Unix to really emphasize portability. Version 7 also formed the new basis for future BSD releases and was more or less the direct ancestor of System V.[/i] So, excuse me... Who is playing catchup here. I do not think it is unix... It seems that Windows is reverting back to history. Next time, check your facts before you make a statement like that. Dan

Neon Samurai
Neon Samurai

The post I responded too seems to indicate that other OS where just now catching up to the Windows platform. I submit that Windows is finally taking baby steps towards becoming a modern OS; moving towards user seporations, requiring third party software to behave better. Yes, mapping to a directory mount point instead of a letter label mount point is not new but in terms of "other OS are finally evolving into a modern OS like Windows"; that is an opinion that is completely backwards. Maybe I read the original post's intent wrong similar to how you seem to have read my comment incorrectly. :)

O & G IT Guy
O & G IT Guy

Neon: Did you miss the posts that indicate that although the original author is showing it in Windows 2008, it has been around since NT? So window's isn't just taking "baby steps" to catch up to this feature that has been around for ages, but rather has had it for ages, and the author is just "rediscovering it".

Neon Samurai
Neon Samurai

then computer science was on to it long, long ago. It's nice that Microsoft is making baby steps to finally catch up too the modern OS that's evolved through posix design principles though. :)

estcst
estcst

The Linux community is a funny bunch like that. They keep acting like they're onto something new and innovative when most of us knew what they were dealing with a couple of decades ago.

mwclarke1
mwclarke1

finally trying to catch up to a real OS

RanchoHamMac
RanchoHamMac

I have been using this in both server and desktop Windows OS's for quite a while. I occaisionally run into a problem where I cannot delete a folder under a mounted drive. Say that I have a physical drive (F:) mounted at C:\Data. And I have a folder called "foo" at C:\Data\Scratch\foo. Windows will not let me execute "del C:\Data\Scratch\foo", but I can execute "del F:\Scratch\foo". I have admin rights to this machine. Go figure.

Neon Samurai
Neon Samurai

In my case, I have this setup: c: c:\Program Files\ c:\Windows .. Program Files containing any installed programs that can not run from there own folder unless installed elsewhere. d: d:\Program Files\ d:\GameFiles\ d:\PalmFiles\ d:\My Documents\ d:\My Workbench\ d:\My Archives\ .. Program Files contains programs that run from there own directory like mIRC and a mirror of my flashdrive's ProgramFiles since all the portableapps run clean. .. GameFiles and PalmFiles contain subdirectories for those respective collections. If I blow away C: on a whim, the games may need reinstalling to replace .dll and registry entries but the save games and addd moduals are all ready and waiting. The same for my PalmOS sync software and various programs (this is a legacy item that will go away when the T5 hits ebay). .. My Documents and My Workbench are short term working storage and long term document storage. My Archive contains any saved installs and drivers of interest; at minimum, what is required to rebuild the Windows install without a network connection or stack of driver disks. It's a natural self backup through drive layout. The next level of "uh crap.." is having to reach for the actual backup DVDs but this setup has saved me many times by simply keeping the storage seporate from the OS build partition. On Unix like systems, it's a natural given. I can reinstall or upgrade my entire OS and loose nothing provided I don't format the /home/ partition with my user account settings and saved data.

epaalx
epaalx

"and the other as Program Files" .. "I just have a small image of my os to restore if need be" I don't see point of having a separate partition for Program Files - you're missing registry and various system files.

Neon Samurai
Neon Samurai

When I started really mucking with Windows and various OS to learn them, I quickly realized that a setup where the OS could be formatted and reinstalled on a whim without loosing the driver installs archive or personal documents was the way to go. I found out about mounting to a directory name in Windows late so I've never bothered or had reason to explore it. I think your setup is even cleaner than my standard C:/D: setup though.

Editor's Picks