Making many network shares appear as one

Givens:

• disk is cheap, and there's plenty lying around on various computers around the corporate intranet
• redundant contiguous large storage volumes are expensive

Problem:

It would be fantastic to have a single entry point (drive letter, network path) that presents all this space as one contiguous filesystem, effectively abstracting the disk and network architecture from the paths presented to users.

Does anyone know how to implement such a solution? I'm open to Windows and non-windows solutions, free and proprietary.

From serverfault jimbojw

• GFS

  GlusterFS

  And many more.

  jimbojw : Thanks Ignacio - Gluster looks promising. I work in a Windows shop, so I'm going to look into DFS first, but these options look like good alternatives as well.
  From Ignacio Vazquez-Abrams

• I believe what you are looking for (on a windows server anyways) is DFS (Distributed File System).

  Read up on it at http://www.microsoft.com/windowsserversystem/dfs/default.mspx

  To fool around with a very basic setup (i did on Windows 2003 SBS):

  • Start > Run > "mmc"
  • click on File > Add/Remove Snap In...
  • Click on the Add button on the dialog that opens and add the "Distributed File System" item.
  • Click OK to add it to the console.
  • Right click "Distributed File System" and choose "New Root" to get started.

  Basically you will create a "Root" UNC path that shares a specific folder and once that's done you can right click the root you just created and click on "New Link" to add additional folders from multiple locations/drives.

  jimbojw : In mmc, I don't see anything labeled "Distributed File System" under Available Snap-ins. I'm on Windows Server 2008.

  jimbojw : Update: After adding the DFS role to my server, "DFS Management" shows up in the Available Snap-ins. Had to create a namespace, and it appears I have more to learn - but thanks for putting me on the path!
  From Make it useful Keep it simple

• You might look at CloudIQ Storage. It allows you to use the drives on commodity based hardware and make it look like a single namespace. A nice feature about it, is that when you write a file, you can specify how many copies you want saved. So if you say 3, the file will be written to 3 machines. If one of your machines crash, no big deal, the file is saved on 2 others. So you are basically getting RAID like functionality on commodity based hardware.

  Its a REST based solution, so you basically perform PUTS and GETS of the files.

  From Brett McCann