minimizing RBrx snapshot size

Discussion in 'backup, imaging & disk mgmt' started by osip, Jul 4, 2007.

Thread Status:
Not open for further replies.
  1. osip

    osip Registered Member

    Joined:
    Oct 25, 2006
    Posts:
    610
    Maybe this has been posted before, nevertheless...I´ve been surprised to see that the daily snapshot size on my wife´s w98 are almost down to zero when in my case on xp pro they in general are on 250-500MB...So,after some more or less heavy activity (deleting,install/uninstall) I made a test snapshot ´X` in win on around 500MB, and then immediately after took a new one with a more appropriate name and deleted ´X´...The new snapshot came out with 0Mb...After a while made a restore and the snapshot was totallyt allright...So, to bring down the snapshot´s size you can try my method...
     
  2. nexstar

    nexstar Registered Member

    Joined:
    Jun 23, 2004
    Posts:
    371
    Location:
    Southampton, UK
    Osip, I may be wrong here but, do you think that RB could be lying about the snapshot sizes?

    I was thinking that the snapshot size is determined at the time it is taken and is most likely based on the incremental changes since the previous snapshot. Now, the displayed snapshot size never changes but even if you delete the preceding snapshot then those files which were linked to it are also linked to the slightly later snapshot and so will remain until all snapshots which are linked to them are deleted.

    That would mean that the later snapshot is potentially now whatever size it was plus the files from the previous (deleted) snapshot.

    Does that make sense? (I've read it twice and I'm still not sure :eek: )

    Graham
     
  3. osip

    osip Registered Member

    Joined:
    Oct 25, 2006
    Posts:
    610
    Every snapshot is unique...no incrementals.They are made only in comparement with the baseline whenever taken...However this size matter is a little bit confusing but nevertheless a fact...(can ´t explain it,try and test it...)
    Another example:made earlier a total scan with KIS:after the snapshot was on ~2.0GB!!!Took a new snapshot and deleted the former...and the size was very small...
     
  4. nexstar

    nexstar Registered Member

    Joined:
    Jun 23, 2004
    Posts:
    371
    Location:
    Southampton, UK
    Well, at least I was right about that ;) .

    I feel that this is going to make my head hurt so I'm going to try some sleep first to see if that helps :) .

    Graham
     
  5. nexstar

    nexstar Registered Member

    Joined:
    Jun 23, 2004
    Posts:
    371
    Location:
    Southampton, UK
    I've tried a few brief tests. I had a large (1.3GB) snapshot on one of my systems which wasn't needed so I deleted it. The free space reported by Windows and by RB was still the same afterwards. The space was only liberated after I ran the RB de-fragger. I guess that's to be expected.

    I did find some relevant info here in one of the faq's on the Horizon DataSys site:

    What is Incremental Sector Redirection?
    Incremental Sector redirection is is what makes Rollback Rx so fast and compact.

    1) Windows operates at a "cluster" level, which consists of multiple disk sectors. Rollback Rx works at the disk sector level.

    2) A snapshot freezes sectors to their current data. Rollback tracks these.

    3) If Windows attempts to update that sector, Rollback grabs the write and redirects it into unprotected Sectors. If Windows attempts to delete that sector, Rollback intercepts the delete. This is "Redirection"

    4) The next snapshot additionally freezes only those sectors that have been written since the last snapshot. This is the "Incremantal" part.


    I'm still not too sure what the snapshot size really means though and just how much of an indicator they are when other snapshots are deleted. Might try some more tests yet :) .

    Graham
     
  6. osip

    osip Registered Member

    Joined:
    Oct 25, 2006
    Posts:
    610
    Interesting...
    Yes, that´s probably some sort of incremental within the current snapshot...could explain that the inteception has occurred after some heavy actions in current and fixed in snapshot 1(still the snsh is made in compare to baseline)...If you then make a new one 2 and deletes no1 the interception is smaller due to almost none action and once again in compare with the baseline...
    Well, just a thought...probably there are guy´s out there who can explain it more adequate...
     
  7. nexstar

    nexstar Registered Member

    Joined:
    Jun 23, 2004
    Posts:
    371
    Location:
    Southampton, UK
    I hope so:) .

    I've just found this in their KB which seems to give a better explanation. I'll have a proper read of it later when I can concentrate o_O .

    Graham
     
Loading...
Thread Status:
Not open for further replies.