Disk size increase with sharing/branching ??

[jkp*74]

I just wonder does folder sharing/branching produce exact copy which occupy the same amount of disk size as the original thus increase the database/disk size dramatically? From the Node Counter tool, it appears to be so.

[Beth]

A share or a branch create another noted in the tree, but that doesn't mean that history or the original file is duplicated.

The best way I can think to explain this is to look at branching. I have File A and I will make a branch of File A and call it Br-FileA.

File A ---- File A ---- File A ---- File A ---- File A ---- File A ----
v1...........v2..........v3.........v4...........v5.........v6
......................................|
......................................|
......................................Br-File A-----Br-File A-----Br-File A-----
.......................................v5..............v6............v7

BR-File A relies on the part of the File A history prior to the branch. When Br-File A is being put together from deltas, it will use Br-File A versions 5-7 and File A versions 1-4. File A versions 1-4 are not duplicated. The node increase you see is the creation of Br-File A version 5. I would normally expect File A v1 to make the biggest increase because that's adding an entire file. Many, many nodes can slow down performance, but that doesn't mean they'll automatically make a large increase in the size of the database.

If you are concerned about the number of nodes you have, then explain your situation further and how you work with your shares and branches. Are you seeing large database increases or taking a hit on performance?

[jkp*74]

Apart from checking in source codes, there are other groups that check in electrical drawing diagrams. Those drawings cause a sharp increase in the folder size (up to 600MB) for certain project. [Note: we are working internally to reduce the file size]

1) New Project A folder created -> Drawing Folder created -> Files add in -> project completed
2) Project A(2) folder created -> share Drawing Folder from Project A to Project A(2) -> continue to add in new files -> project completed
3) Project A(n) folder created -> share Drawing Folder from Project A to Project A(n) -> continue to add in new files -> project completed

Can i say that if Project A/Drawing Folder occupy 600MB, then Project A(2) / Drawing Folder also occupy 600MB after the sharing operation?

Reason for sharing the Drawing Folder is for easy retrieval of those diagrams correspondent to a specific Project A(x).
I am more concern over disk size than performance at this moment.

[AjarnMark]

Joel, this may not be your situation, but thought I'd throw out our experience that sounded similar in case it helped.

Within my company, we had some confusion in communication where some people were discussing amount of disk space consumed in their working folder due to branches and shares, and confusing that with the amount of space consumed in the repository database. Within the repository, sharing and branching are relatively lightweight operations because of the common history and storage of changeset information rather than full copies in the repository. However, when you GET all that to your working folder, each one of those shares or branches results in a full copy of the file in your working area.

So, in our case, some folks' local hard drives were getting filled up with the old branches and shares that were no longer actively edited, and they mistakenly thought it meant that the repository was just as physically large, but it is not.

[Beth]

AjarnMark: What you are saying makes sense. When you perform a get of everything, you do use up more local disk space.

jkp*74: If your files are images, then what could be happening is that every changes results in a huge delta, because usually 1 change affects all the rest of the file to the point where the entire thing has the potential to look completely different at the binary level. Large deltas will definitely contribute to a database size increase. It's not so much the branch's fault, as it's still doing its usual thing, but if you have two paths creating large deltas, that increases the database size twice as fast from the branch point forward. I wouldn't expect share to hurt things much as the delta won't get uploaded twice. That becomes just an issue in the number of tree nodes you make.

[jkp*74]

Dear AjarnMark,
Thanks for the explanation. It does relief some of my worries.
About the increase of local hard drive size, it does happen to the staffs here.

Dear Beth,
Fortunately those drawings won't go thru many changes after the branch operation.
Thanks for the explanation.

[Beth]

Thanks for the update. If you have any further questions, just let me know.