4

How do distributed filesystems differ from cluster filesystems?

Is it just the wording that I am getting tripped up on? Are there any significant differences between the two.

2 Answers 2

4

Technically, Distributed filesystems is a file system that can be accessed anywhere on a network. So a Network Filesystem (or NFS), is a distributed filesystem.

A cluster filesystem is a filesystem which is mounted on multiple devices, keyword is mounted. So physically, it does not need to be on the network, etc.

So in this case, a cluster filesystem is a distributed filesystem, but a distributed filesystem "can" be a cluster filesystem.

6
  • Can you sort of clarify what the part about cluster fs? I don't see what the purpose would be if they are not connected to a network (in that case they'd just be different computers with identical files)? Thanks :) Nov 2, 2011 at 1:13
  • A clustered FS is able to be mounted on multiple computers. For example, a drive may have multi USB support, but yes. Most of the time it is done over the network. By mounting, I mean it's virtually like an attached disk to the desktop.
    – Steven Lu
    Nov 2, 2011 at 1:49
  • So basically for the cluster fs, if a change is made to a file, will all computers see it immediately after the change is made? Nov 2, 2011 at 6:02
  • 1
    Yes, it should be immediate. But then again, there are some overhead things that each system has to do, so not instantaneously, but rather, very quickly and immediate.
    – Steven Lu
    Nov 2, 2011 at 17:13
  • Ah, I see now! Which one is better then? It seems like a distributed file system would allow you to avoid the overhead of writing to each computer's disk. But then again, in a distributed model, if that one disk with all the files goes bad, then you're screwed XD Nov 2, 2011 at 19:15
1

A distributed file system does not mean it will fail on one disk dying. A distributed file system can be a clustered FS if the distributed FS is mounted to the system.

An FS can be replicated such as in RAID to avoid data loss if a hardware disk ever does die.

For example, if I wanted one server to serve a dynamic file located on the FS, I can do so through multiple servers, instead of one. That would be one reason I'd use distributed file system.

2
  • I still don't understand what you mean there's a difference in how they're mounted...? I realize I've asked a lot of questions, but I still don't get it xD Nov 3, 2011 at 3:25
  • Okay, a distributed file system is a file on the network. A clustered file system is any file system that is "mounted" to multiple computers. So from the definition above, you can see that a distributed file system can be a clustered file system because they can be mounted on multiple machines. The reason we use this is so that a cluster of machines can serve the same data (which is on the) clustered filesystem (which can also be a distributed file system). Note that a file system can be redundant, avoiding data loss if a "disk" dies.
    – Steven Lu
    Nov 3, 2011 at 7:49

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .