none
Help me design my DPM RRS feed

  • Question

  • Get any coffee or bathroom breaks out of the way before starting to read this post J

    I have installed DPM 2012 and am evaluating this product to replace my Backup Exec 2010 system, which I have been using for several years. We recently implemented Hyper-V on a two node, Windows 2012 cluster, and many of my servers have already been virtualized. I have about 15 servers running everything from Windows 2000 to Windows 2008 R2, although most of them are Windows 2003 by now. I’m planning to take local backups of my Win2000 machines and just save those to one of my 2003 boxes and back them up from there. My recovery goals basically break down into 3 areas. Below I list each of these and then I explain how I am considering addressing each one. Much of what I say below is based on a very basic knowledge of DPM, I have only been using DPM for a few weeks and am just now beginning to understand the basics.

    1…File Shares : I need 2-3 recovery points each day Monday through Saturday and keep these backups for 7 years. Users need to recover files themselves.

    Create a Protection Group called “File Shares”

    Group Members: File shares only, no system state other data included

    Short Term Retention: 21 days

    Synchronization: 2 hours

    Recovery Points: 11am and 8pm

    Disk Allocation: 5TB

     Long Term Retention (tape): 90 days (Each month a full backup is archived, these are kept for 7 years)

    2…Disaster Recovery : I need full disaster recovery points taken each day, for all servers, and kept for only 2 weeks.

    Create a Protection Group called “ Disaster Recovery”

    Group Members: VHDs of All file servers, application servers, Exchange servers, Domain controllers

    Short Term Retention: 3 days

    Synchronization: Before creating Recovery Point

    Recovery Points: 8pm

    Disk Allocation: 3TB

    Long Term Retention (tape): 14 days

    3…Exchange and Active Directory Item Level Recovery : I want the ability to recover any item that has been deleted within the last 30 days. Both Exchange 2010 and AD 2008 include item recovery via their "recycle bin" features. DPM is not needed to address ‘item level recovery’ for either Exchange or AD.

    Questions:

    1. For my file shares, does it really make sense the way I have made a specific protection group for this? Some people on the forums say that you cant have the same server show up in two different protection groups, is that true? I hear that because I am also doing VHD backups, that I will essentially be storing twice as much data because I will store both a full VHD replica and a replica of the File Share. Is this true? Perhaps a better idea would be to have a Protection Group for all my file servers and to include all file shares AND disaster recovery type data. Not sure if I could do that and still provide EUR.
    2. Am I understanding correctly, that synchronization only affects the frequency with which the backup data is refreshed, it does not affect the size of the backup data being stored. In other words, synchronization frequency is not one of the considerations when sizing DPM storage.
    3. I do not understand how the recovery point schedule impacts DPM storage requirements. What takes more space, one recovery point or 5 recovery points for the same 24 hour day? I understand we only get 64 recovery points for any server volume, but I don’t understand if increasing the frequency of recovery points will result in less storage consumed or more.
    4.  When creating my first protection group, I added server1\share1. DPM prompted me to specify disk allocation parameters or to accept the defaults and continue. However, when I go back into the same protection group and I add server1\share2, DPM does not indicate in any way that disk allocation parameters need to be adjusted or even reviewed. How can that be? Adding another share should mean that I will need to adjust disk allocation. So on my own, I go to Disk Allocation and click ‘calculate’ and the number does not change. Why? This number should change based on how much data is being selected, correct?
    5. For my Disaster Recovery needs, it seems simplest to just take snapshots of my VHDs every day. Then when restoring, all I do is restore the VHD and recovery the latest recovery point from my ‘File Shares’ protection group and the server is back. Is this reasonable or am I missing something big by not utilizing ‘System State’ backup or BMR. Does a VHD backup work well when recovering a failed Exchange server or a failed domain controller?
    6. For recovering items in Exchange or Active Directory, do I need to use DPM if the same type of functionality seems to be available in Exchange and AD themselves?
    7. Are my questions so bad that I have proven how dangerous I really am and I should just slowly back away from the keyboard?
    Thursday, May 2, 2013 9:54 PM

Answers

  • 1. I don't have End User Recovery enabled, but I don't believe that Hyper-V backups allow for end user recovery. 

    2. Synchronizations x-fer the block level changes provided by the DPMRA agent every time you synchronize - so it is all dependent on change rate. The total change is kept until a Recovery Point job is run at which point all those synchronizations are finalized as changes to the Replica Volume.

    3. You may end up with some slight overhead in files that change in between every single recovery point, but the key is the change rate. If you change 1GB in between every single Recovery point, its going to keep track of that 1GB over the total retention period. Yet, if you cut the recovery points frequency down, and then there is now 2GB of change in between your recovery points - its still the same amount of change (same amount of disk), and you have half the recovery points to recover from.

    Possible drawbacks is that you won't be able to keep very many days as you will max out the max number possible in a short period of time.

    The point of recovery points only twice a day, or 5 times a day is for specifying a recovery model to the limitations of VSS.  You end up with a maximum number of recovery points -- so if you want 8 per day.. you could do 8 per day for 7 days.  Or.. 1 per day for 64 days. Its the 64 snapshot ((sorry, I forget the specific number)) that will limit what you can do with VSS - FULL COPY snapshots.  The other point here is that application services such as backing up SQL DBs, Exchange, or SharePoint directly are handles slightly different depending on recovery models.  -- For a SQL DB set to Full -- every synchronization is an incremental snapshot -- in which case the VSS limitations bump up to 512. 

    Also, since you are new to DPM - I highly advise reading through how short term disk/tape and long term tape retention works.

    Short Term Disk = Retention period as specified. Oldest recovery points are removed from disk as the retention period is hit - so your 22 day retention period. as day 23 hit, day 1 is removed from disk.

    Short Term Tape = 1 Full Replica is written to a new tape for every recovery point. Synchronizations are written to an additional Tape , and then appeneded until a new recovery point.

    Long Term Tape = 1 Full Replica is written off to tape - point in time.... so if you are looking to keep monthly tape protection, When you go to restore it - you can restore 1 Recovery point from that entire Month.

    So.. 22 days on disk is only ever going to be 22 days on disk. You won't be able to keep that for 7 years.  

    Writing off to tape, you will keep 1 recovery point from 1 day per month. If you look to write more recovery points off to tape. Keep in mind that DPM does not append recovery points on long term storage. It will use new Tape for every long term tape job -

    • Marked as answer by tallcann Friday, August 2, 2013 9:42 PM
    Friday, May 3, 2013 5:40 PM

All replies

  • DPM protection groups are aware of the volume ID, and use VSS writers on that particular volume or mount point. The limitation of splitting data from the same server to multiple protection groups is dependent on that data living on different volumes.  Example: Volume 1 (c:\) system state goes to DR protection group. Volume 2 (d:\somefolder) goes to your file share protection group. That is OK, but if you want to include c:\someFolder to a protection group, it would have to be in the same protection group as the anything else being protection from that volume. The exception to this rule, is if you have broken off parts of the volume using MountPoints. The file system will recognize the mount point as a different volume.

    As to VHD protection. IF you have Hyper-V enabled on your DPM server, and you protect a VM through the host's Hyper-Visor. You CAN restore file level from inside the VHD. So yes, backing up the VHD from the host and the Files from the VM is wasting space. This requires a physical DPM server.

    Recovery points... look at how the data is stored. You have a Replica and Recovery point volume for every protected server. The Replica volume is a mirror of the protected data -- it will always depend exactly on how much data you are protecting. If you are protecting 200GB of data, you will need to allocate enough space to accommodate that. The Recovery point volume contains the differential vss data. This is impacted on how much data changes and how long the the recovery points are kept. Example, I have a file server that actively changes about 200GB per day, and keeps the twice daily recovery points for 1 week.  7 Days * 200GB/Day of CHANGE = roughly 1.4TB of data in the Recovery Point volume.  Its all about the daily change rate -- which you may not know right away.

    5. System State and BMR are not the best thing to use when dealing with VMs - taking a snapshot of a virtual machine accomplishes the same thing.  -- system state and BMR would require recovering the image to a network source, booting hardware/vm to windows recovery, and restoring the image.

    6. Item level recovery isn't possible, but as a DR stand point - it does make sense. AD Recovery Bin is a pain with powershell, but I suppose is now available as a GUI w/ 2012. But there are scenarios where AD needs to be authoritatively restored in its entirety - in which case I hope you have a System State / Sysvol backup from your domain.

    Hope that helps some

    Regards,

    Kyle

    Friday, May 3, 2013 3:26 PM
  • Thank you Kyle, your answer was very helpful.

    1. So perhaps a more efficient way to backup my file servers is to create a "File Servers" protection group and target the VMs of all my file servers through Hyper-V. This will provide item level recovery, BUT will this still allow "End User Recovery" from the Windows Explorer "Previous Versions" tab?

    2. Synchronization schedule does not affect the space required for storing backups, correct?

    3. My question was referring to *frequency* of recovery points. So to use your example...What if you still stored 7 days of recovery points, but instead took recovery points 5 times per day instead of 2. The size required to store this would still be 1.4TB, correct? If so, what would be the point of taking recovery points only twice a day? If taking recovery points every hour results in the same data stored, then what are the drawbacks of doing so?

    For questions 4-7 I'm good, thank you!

    Friday, May 3, 2013 4:45 PM
  • 1. I don't have End User Recovery enabled, but I don't believe that Hyper-V backups allow for end user recovery. 

    2. Synchronizations x-fer the block level changes provided by the DPMRA agent every time you synchronize - so it is all dependent on change rate. The total change is kept until a Recovery Point job is run at which point all those synchronizations are finalized as changes to the Replica Volume.

    3. You may end up with some slight overhead in files that change in between every single recovery point, but the key is the change rate. If you change 1GB in between every single Recovery point, its going to keep track of that 1GB over the total retention period. Yet, if you cut the recovery points frequency down, and then there is now 2GB of change in between your recovery points - its still the same amount of change (same amount of disk), and you have half the recovery points to recover from.

    Possible drawbacks is that you won't be able to keep very many days as you will max out the max number possible in a short period of time.

    The point of recovery points only twice a day, or 5 times a day is for specifying a recovery model to the limitations of VSS.  You end up with a maximum number of recovery points -- so if you want 8 per day.. you could do 8 per day for 7 days.  Or.. 1 per day for 64 days. Its the 64 snapshot ((sorry, I forget the specific number)) that will limit what you can do with VSS - FULL COPY snapshots.  The other point here is that application services such as backing up SQL DBs, Exchange, or SharePoint directly are handles slightly different depending on recovery models.  -- For a SQL DB set to Full -- every synchronization is an incremental snapshot -- in which case the VSS limitations bump up to 512. 

    Also, since you are new to DPM - I highly advise reading through how short term disk/tape and long term tape retention works.

    Short Term Disk = Retention period as specified. Oldest recovery points are removed from disk as the retention period is hit - so your 22 day retention period. as day 23 hit, day 1 is removed from disk.

    Short Term Tape = 1 Full Replica is written to a new tape for every recovery point. Synchronizations are written to an additional Tape , and then appeneded until a new recovery point.

    Long Term Tape = 1 Full Replica is written off to tape - point in time.... so if you are looking to keep monthly tape protection, When you go to restore it - you can restore 1 Recovery point from that entire Month.

    So.. 22 days on disk is only ever going to be 22 days on disk. You won't be able to keep that for 7 years.  

    Writing off to tape, you will keep 1 recovery point from 1 day per month. If you look to write more recovery points off to tape. Keep in mind that DPM does not append recovery points on long term storage. It will use new Tape for every long term tape job -

    • Marked as answer by tallcann Friday, August 2, 2013 9:42 PM
    Friday, May 3, 2013 5:40 PM
  • Okay so let's see if I understand what you're saying...

    1. MAYBE. Need to verify whether EUR is available from a Hyper-V level backup.

    2. INCORRECT. Synchronization schedule DOES affect space required for backup storage because each sync is kept until the next recovery point job.

    3. CORRECT. Space required for backup storage does not change based on frequency of recovery points, so the only consideration for deciding how often to run recovery points is that there is a limit to the number of recovery points I can keep.

    ...How did I do?




    • Edited by tallcann Friday, May 17, 2013 10:24 PM
    Friday, May 17, 2013 10:22 PM