About Vertica Backups through vbr

Question 1:

Can we run Multiple backups (object level) same time in multi node cluster That means In Node1, I launched the vbr.py for Table1 , same time can I able launch vbr.py at Node2 for Table2.


Can I able run multiple backups parallelly in a multi node cluster (KSAFETY 1 environment) through vbr.py.


Question 2:

What all are factors will impact to slowdown the backup process?

In what scenarios, Backup performance will degrade. 


Question 3:

 If i took the schema level backup (one ini file), Is it possible for me to restore only one table from that backup (ini file) 



  • Options

    What's the Vertica version ?

  • Options



     Vertica Analytic Database v7.2.3-7

  • Options

    1: No, you can run only one backup task with the same temp dir, snapshot name and backup locations, regardless of the node you are launching the vbr.py.


    2: You are using one of the latest releases, unless you are still having the old storage layout, there is nothing from Vertica side that could slow down the process. Initial backup takes more time. Performance of subsequent backups must be propotional to the database churn, how much storage is mutated since the last backup as Vertica only transfers deltas. You still have a fully consistent backup to restore at every backup point.


    Obviously other factors, such as network (between Vertica nodes and backup hosts) and disk IO (probably backup hosts) could affect the speed. Customers usually take backups during less busy hours, so it doesn't compete with disk IO for normal workloads. We've seen some had network shaping in place that affected backup performance. 


    Old storage layout is what we had prior to 7.2, which could create millions of files. This can greately reduce the backup performance. 7.2 is much better (10x or more) than 7.1, but small files can still hurt.


    3: Yes. this is called subset restore. You can take any backup (database, schema or tables) and select a subset of objects (schemas or tables) to restore.  The command line option you want to use is "--restore-objects=<names of subset of tables/schemas to restore>". Find more details in the following link.



Leave a Comment

BoldItalicStrikethroughOrdered listUnordered list
Align leftAlign centerAlign rightToggle HTML viewToggle full pageToggle lights
Drop image/file