Incremental Backup and Restore with Productstream

• 

  Brian Schanen joined Autodesk in 2005 as a Product Designer and currently is a Customer Success Engineer for Autodesk’s Data Management products. Brian has taught at Autodesk University numerous times and has authored white papers on Vault and Productstream. He works with customers to implement a complete digital prototyping solution specializing in Autodesk Inventor, Autodesk Vault and Productstream. Brian lives near Detroit, Michigan.

  About Brian

Latest Post

• Add to Bookmarks Send to Peer

  Incremental Backup and Restore with Productstream
  September 8, 2008, 09:57 PM Brian Schanen

  Incremental Backup is a server side feature of Productstream and the conventional wisdom has been that you need to restore sequentially beginning with the last Full backup and then every Incremental in Sequence. For Restoring with Productstream 2009 you can simply point to the last Incremental Backup and it will automatically find the remaining data - provided all data is retained and not moved around.

  Note: Always use the built-in ADMS Backup and Restore functionality, to ensure synchronization between SQL database and Filestore.

  The benefit is obviously that a Restore is faster and needs less interaction, as it will Restore the latest Database and the Incremental Filestore Changes, from the latest Full Backup and Subsequent Incremental Backups.

  An Incremental Backup consists of a Full Backup with a following series of Incremental Backups. Each Incr. Backup consists of a Full Backup of the (SQL) Database plus just the changes to the Filestore (usually addition of files but also information of deletions and purges of file).

  The functionality relies on retaining the Backup Folder structure – meaning all backups (Full and Incremental) should exist on the same level.

  How to Use:
  This example is based on a Full Backup on a Sunday and then 5 Incr. Backups on each of the following days.  This will mean that you should run 2 different Schedules Backup Tasks – a Full Backup on Sundays – and then run another Scheduled Task with Incremental Backup the following 5 days. The only difference is the command line option “-INCR“ (see Help on “Commandline“ in ADMS)

  Before a Restore - all backup sets need to be present in same subfolder (In the example C:\Backup)
  In Example below all Backups are taken on the same date – a little cheating!

  

  The naming format for folder name is VaultBackup_<year>_<month>_<day>_<hour>_<minute>_<second>_<AM or PM>_<Full or Incr##>

  When restoring point to the LAST incremental backup folder when asked for location (Tip: Sort on date or look for highest Incr## number). It also works if you on Friday want to get back to Wednesday – just run Restore from ...Incr03 folder.

  How does it work:
  In every Incr. folder exists a file “BackupHistory.txt“.

  

  This contains the names of the Full and Incremental backup data set from last full backup. This information is used by the Restore function to find the names of the previous data sets.

  

  To use this functionality, a little care might need to be taken in thinking ahead on the “What if ...“ scenario the whole Backup/Restore process is intended to overcome. If the unthinkable should happen and the server dies, and you have to setup a new machine (better would be to have a test server ready to run), then there is probably too much to think about to remember to add Content Center Libraries for Inventor afterwards. In that light it is much easier just to include the Libraries in the Backup (even if they take some space), as you would not need to search for the Inventor DVD, while restoring the Database.

  Finally we have added a few comments and tips related to Backup, as they could be beneficial if you consider modifying your scripts and procedures anyway:

  • This was stated before (above) but here it is again: Always use the built-in ADMS Backup and Restore functionality, to ensure synchronization between SQL database and Filestore. With incremental backup time spent is minimal, and it ensure a painless restore (if needed later which - everybody hopes not). There are LiveBackup options from different Backup providers, but they leave You with the head ache of ensuring the timing of the 2 components.
  • Restoring a Backup with no CC Libraries (-DBSC) will still remove existing Libraries
  • Note that if you have Productstream 2008 a detach/attach invalidates the Backup set from Restore – not so anymore in Productstream 2009. The popular approach of “cycling“ backup sets from A to B folders should not be used any more, as Incremental Restore relies on consistent location to find the datasets needed for a successful restore.
  • Backup will only run if data is Replicated first. If not it will set off a Replication first. If a database is to be Restored – do it on the location closest to the SQL server for performance, and you will need to re-enable the Vault on the remote sites (and point to the existing remote Filestore).
  • Remember that with Productstream 2009, hot backup allows users to keep working while backup is running.
  • There is no need to stop/start IIS or SQL server before/after Backup anymore. Maybe prudent still to do on Sunday after the SQL Maintenance Plan has run. See Advanced Configuration Guide for setting up a SQL Maintenance Plan – highly recommended if not already done.
  • The old –defragment option has now fragmented (Sorry!) into 2 options “–defragmentVault“ and “-defragmentLibrary“. Good to use after Backup – but stops users working while running.
  • Regarding the timing of the Backup. It might be good to check with the White Paper “Vault and Productstream Background Processes“ as the ADMS will be busy with some operations at specific time slots http://images.autodesk.com/adsk/files/vault_background_processes_whitepaper.pdf .

  We hope these options have been beneficial, and hope you will comment on the usage.

  Special thanks to Uffe for the detail in this post!

  -Brian Schanen

   

   

   

  0 Comment | Add Comment Under The Hood >

Comments