how does Retrospect save Data?

[kruessi80]

Hi,

I'd like to understand how Retrospect works.

 

We use 4 running Units "Ausführungseinheiten".

To each I have one Backup-Set which backups the Client-Data to 1 big hard drive.

I have to backup Mac's and Windows 2000 and Windows XP PCs.

Now, I have mixed these three types of Maschines in the Backup-Sets and these are very large.

(about 170GB)

 

Will the Backup-Sets get smaller, if I have one Backup-set to backup the Mac's, one

Backup-Set to backup the Windows 2000 clients and one Backup-Set

to backup the Windows XP clients?

 

(we also do an copy of each Backup-Set to one Backup-Set on DAT-tape.

There are only 90GB used. So there's my Problem/Question.)

 

------------

 

If Client 1 and Client 2 have the same "test.txt"

Is it like I.) or II.)

 

I.)

Client 1 "test.txt" ->

_________________ -> Retrospect -> Backup-Set -> "test.txt"

Client 2 "test.txt" ->

 

II.)

Client 1 "test.txt" ->

_________________ -> Retrospect -> Backup-Set -> "test.txt"(client1), "test.txt"(client2)

Client 2 "test.txt" ->

-------

 

Sorry for my bad english...

 

greetings

Matthias

[nekr0phage]

Hello Matthias,

 

Retrospect compares files on a source volume against the entire backup set. The same file located on multiple machines will only be backed up from one machine, but each machine's snapshot will have a map to that file.

 

You must remember that Modification Date and Time are matching criteria for Retrospect. So, while many common files like Windows OS files will likely only need to be backed up once per backup set, any data files that have been copied across volumes will likely get copied multiple times to compensate for changes to the modification date/time.

 

Hope that helps.

[willflint]

Quote:

---

 

Retrospect compares files on a source volume against the entire backup set. The same file located on multiple machines will only be backed up from one machine, but each machine's snapshot will have a map to that file.

 

---

 

Is this documented anywhere. I have another posting on the forum whereas I am backing up a server with over 4.5Million objects. Due to one session having more than 4M files, Retro would crash. A workaround I am trying is to backup the 100 subfolders that contain 98% of the files as individual subvolumes into 2 backup sets. That is working (so far) to alleviate the 4M per session issue, but as far as I knew, if the files moved between the folders (which it appears that they do, and the moves are done by a database) then I would get multiple backups of singular files. But according to your posting, that may not be case (barring the file's modified date is not altered). Let me know if you got that info from the documentation, it would be greatly appreciated.