Folder on network: "size on disk" is much smaller than "size" Why?

Factual Questions
1

So at my work, I wanted to see the size of one of our network folders. It is a folder that multiple employees can access on some server somewhere.

So I right clicked and went to properties:

Size: 130 GB

Size on Disk: 1.87 GB

Now I am aware that sometimes “size on disk” can be larger if you have many many small files because of large cluster size. This is obviously not that problem.

I am also aware you can have a smaller “size on disk” due to compression. But I know that those 130 GB are not compressed down to less than 2 GB. I know there are multiple (already compressed) video clips on there that are several GB.

Why is the “size on disk” so wildly different? Is this just a bug/fluke of network drives?

2

It could be that some of the files are just links to the same actual copy.

3

That would have to be some trickery the servers and IT personnel do behind our backs. I have personally dumped more than a dozen GB directly into this folder, and it still appears to be there.

4

Might be compressed - so the real size is 130G, but only uses 1.8 on disk.

5

ETA: No, that was just stupid…

6

Could be a buggy fileserver, either under-reporting the size on disk or perhaps reporting it as a 32-bit int (i.e. modulo 2GB).

7

Or

It could be because of drive mapping … the file server can hide the fact that there is separate file systems but the folder usage function is passed over to the server, which then doesn’t know if the desired usage is for the purpose of taking a copy (where you’d prefer to know about the whole tree… ) or solving space problems on the file server (where you want to know how much is on each filesystem… )

8

This was my guess as well. So I tested it on my home network. I couldn’t get symbolic links to work, but a junction point did. However, I could noot reproduce the phenomenon. The parent folder still is at 3.39 GB total and 3.39 GB on disk. And the junction point is 228GB and 222GB on disk, the same as if I check the drive directly.

This was just using Windows 7 on both computers, and was not on a domain. (I don’t use Homegroups, however.)

9

From the hints it doesn’t seem to be your case, but all-zero blocks need not actually occupy space on the disk. They’ll show up in file-length, but not in folder utilization.

(OS doesn’t check for all-zero data and then suppress file allocation; instead these sparse files are created deliberately, e.g. with truncate command. I don’t know what applications do this, but imaging software which knows what the final size of a file will be might find it convenient.)

10

If this is a Windows network, DFS - Distributed File System - might be the culprit. DFS means that different parts of the filesystem can be on different servers. This is very similar to the junction points others have mentioned.

11

A long shot - but if the file server uses a de-dupe capability, and the documents stored on it are lots of versions of substantially the same thing, or indeed actual duplicates, de-dupe will wipe out all the common components. This could be enough to get such an apparent drop in disk use.

De-dupe can extend right across the entire server system, so it will look for common elements in files anywhere. The really big saving is in things like emails that go to everyone - it will wipe out all bar one copy on the server, or common documents that many people have, again, only one copy being effectively stored.