none
what type of backup DPM takes in recovery point volume RRS feed

  • Question

  • Hi,

    Please help me to understand what type of backup DPM takes in recovery point volume as it takes more disk space as compare to replica point volume, though replica point volume is initial full copy of data source and why it takes more disk space.

    Thanks in advance.

    Tuesday, February 4, 2014 12:38 PM

All replies

  • Replica volume keeps an up to date copy of the data backed up.

    Recovery points allows to revert back to a specific point in time.

    Recovery points volume contains all changes between first backup(or actual time - retention time) and actual time. That's why it can be bigger than replica's volume.

    For exemple if you have 15% changes on a file server and a retention of 10 days then recovery point volume would contain ~50% more data than the replica.

    Tuesday, February 4, 2014 1:33 PM
  • So as per my understanding –

    1.        Initially DPM backs up complete data and keep it in replica volume with empty recovery point volume.
    2.        During every synchronization frequency, DPM backs up changed data and transfer it into recovery point volume without touching replica point volume
    3.        During every express full, DPM syncs the data in replica volume with current state of production data and backs up only changed data to replica volume from previous express full and keeps the shadow copy of previous express full in recovery point volume.
    4.        Same process continues till the applied retention range
    5.        At expiry of retention range, whole data from both volumes gets deleted from disk and new data gets backed up on next express full.

    So what I understand is as recovery point volume keeps the changed data as per sync frequency plus all volume shadow copies of replica volume, it takes more space as compare to replica volume which carry only latest copy of data.

    Please correct if I am wrong.

    Also want to understand logic behind disk space calculation formulas mentioned on below link. For example why there is multiple of 4 or 2.5 for exchange or SQL recovery point calculation resp.? Why there is 1600 MB is getting added for each recovery point calculation?

    http://technet.microsoft.com/en-us/library/hh758007.aspx

    Please help me to understand.

    Tuesday, February 4, 2014 4:13 PM
  • Hi,

    your understanding is correct.

    The replica volume evolves at each full backup.

    For the 5th point, at expiration of the retention range there's only the old recovery point out of range that will be deleted from the volume. Replica is kept untouched as well as the RP in the retention range.

    For the calcula you should ask Microsoft directly.

    Tuesday, February 4, 2014 7:05 PM
  • Thanks for valuable inputs.

    But I have still doubt on my understanding i.e.

    I have calculated disk space required as per Microsoft formulas given on above mentioned link. Below table is there output.

    Application

    Data   Size

    Retention   Range in Days

    Replica   Volume

    Recovery   Point Volume

    Total   Volume

    Exchange   Mailbox

    700   GB

    7

    906   GB

    1962   GB

    2868   GB

    SQL   Database

    100   GB

    7

    125   GB

    107   GB

    232   GB

    File   & Folders

    500   GB

    7

    750   GB

    72   GB

    822   GB

    Virtual   Machines

    700   GB

    7

    1050   GB

    100   GB

    1150   GB

    So as we are considering during sync frequency DPM transfers changed data to recovery point volume plus volume shadow copy of replica during each express full for 7 days (as per above table) then how/why required disk space for recovery point volume is less as compare to replica point volume for SQL, File, VM apps?

    Please help me to understand.

    Does anyone else have an idea regarding the formulas given by MS to calculate required disk space?

    Please help.
    Wednesday, February 5, 2014 2:37 AM
  • MS Formulas take into account the % of change for each day and DPM is able to compress some data.
    Wednesday, February 5, 2014 8:10 AM
  • How it relates to above query i.e.

    how/why required disk space for recovery point volume is less as compare to replica point volume for SQL, File, VM apps?

    Please help me to understand.

    Wednesday, February 5, 2014 8:58 AM
  • For exemple in the VM Case:

    The formula is the following (Data source size x retention range in days x 0.02) + 1600 MB

    If you take a look at it you can see a "X 0.02" this means that the formula is taking 2% of change per day.

    If your VM is 10GB then each day you will have 200MB of change. Over 7 days that would be 1.4GB for exemple.

    Apply it to the formula that would make: 10GB * 7 * 0.02 + 1600MB = 3GB

    Repeat for all data source.

    Wednesday, February 5, 2014 9:19 AM
  • So as per that whether formula is considering 200% change for file server? As its formulas is

    File server recovery point = (Data source size x retention range in days x 2) / 100 + 1600 MB

    Also what can be the mean of 2.5 in SQL RR formulas though it contain the parameter called log change

    SQL recovery point = 2.5 x retention range in days x log change x data source + 1600 MB

    Also why there is addition of 1600 MB?

    Wednesday, February 5, 2014 10:06 AM
  • For the file server recovery point that's not 200% but 2/100 so 2%

    These are Microsoft formulas to calculate the size needed but you can override this setting and set what you want during the Protection Group wizard.

    As i told you, you should ask Microsoft directly about these formulas. It is based on their knowledge and experience on this product.

    All these only refers to the default allocation and allows DPM to have disk space available and not expending the volume at each backup as it would consume a lot of ressource.

    Wednesday, February 5, 2014 10:16 AM
  • Thanks Jeremy for your valuable inputs and time.
    Wednesday, February 5, 2014 10:31 AM
  • Hi Parag_M

    I would like to correct your understanding on how recovery points work.

    <snip>
    1.  Initially DPM backs up complete data and keep it in replica volume with empty recovery point volume.

    MS - Correct.

    2.  During every synchronization frequency, DPM backs up changed data and transfer it into recovery point volume without touching replica point volume

    MS - Incorrect, just the opposite, DPM synchronizations always and only write to the replica volume. VSS shadow copy driver (volsnap) performs a copy-on-write (COW) to preserve the old blocks before they get overwritten by DPM.   The COW always and only writes to the recovery point volume.

    3.  During every express full, DPM syncs the data in replica volume with current state of production data and backs up only changed data to replica volume from previous express full and keeps the shadow copy of previous express full in recovery point volume.

    MS - Incorrect - An express full does the exact same as a synchronizations, the only difference is DPM is transferring over only changed blocks of DB files in the case SQL and Exchange, or VHD's in the case of VM's and applying them to the files on the replica.  COW is again being done as blocks are overwritten during express full.

    4.  Same process continues till the applied retention range.

    MS - Correct.

    5.  At expiry of retention range, whole data from both volumes gets deleted from disk and new data gets backed up on next express full.

    MS - Incorrect, every night DPM deletes the oldest snapshot(s) from the recovery point volume which exceed the retention range which makes room for new snapshots when express full or recovery points are made.

    So what I understand is as recovery point volume keeps the changed data as per sync frequency plus all volume shadow copies of replica volume, it takes more space as compare to replica volume which carry only latest copy of data.

    MS - Kind of - It only holds VSS shadow copies of the replica.  The last snapshot taken (after express full or recovery point) is active, and will be written to by COW during incremental syncs and express full backups.  When the next snapshot is taken, the pervious one becomes a read only snapshot and will be deleted after the retention period is reached.
    >snip<



    Please remember to click “Mark as Answer” on the post that helps you, and to click “Unmark as Answer” if a marked post does not actually answer your question. This can be beneficial to other community members reading the thread. Regards, Mike J. [MSFT] This posting is provided "AS IS" with no warranties, and confers no rights.

    Thursday, February 6, 2014 12:35 AM
    Moderator
  • Thanks Mike for detailed explanations.

    Like above explanations can you please tell me the logic behind the parameter mentioned in below formulas of recovery point volume, especially behind 4.0, 2.5 & 1600 MB.

    For Exchange data:

    • 4.0 x retention range in days x log change x data source size + 1600 MB

    For SQL Server data:

    • 2.5 x retention range in days x log change x data source size + 1600 MB
    Friday, February 7, 2014 7:16 AM
  • HI,

    Afraid not, I didn't write the formulas, but they seem to work.

    You can download the DPM Storage calculators which may be more accurate based on other parameters.


    Please remember to click “Mark as Answer” on the post that helps you, and to click “Unmark as Answer” if a marked post does not actually answer your question. This can be beneficial to other community members reading the thread. Regards, Mike J. [MSFT] This posting is provided "AS IS" with no warranties, and confers no rights.

    Friday, February 7, 2014 2:23 PM
    Moderator
  • It’s quite difficult to understand these calculators so I will request if possible can you please explain above formulas.

    Sunday, February 9, 2014 5:27 PM