Forum Discussion

LorisM's avatar
LorisM
Level 3
13 years ago

Overlapping dedupe jobs

Hi,

Running Backup Exec 2010 R3 with dedupe option.

When scheduling backup jobs is it better not to overlap them when backing up to a dedupe folder?

Will overlapping jobs mean im not getting proper deduping of backups if the overlapping jobs have the same sort of content?

 

Im after some recommendations on how I should be scheduling backups to a dedupe folder and then optimize dedupe to another media server. At the moment I am duplicating to another media server at completion of each backup job automatically, is this the best thing to do?


Thanks

  • Definitely, there is a lot more processing going before the data is stored in the dedup folder.  BE has to hash the data and check each block against existing blocks to make sure that it is unique before storing it. For B2D folders, it is just a straight write.

    When you run simultaneous jobs, you have to balance the load on your machine with the time gain.  With a not so poweful machine, you might be better off running consecutive jobs.

8 Replies

  • There is no harm in scheduling your duplicate jobs immediately after your backup jobs.  Only the changed data blocks would be sent to the other server.

  • There is no harm in that either.  Only one copy a data block will be kept in the dedup folder.  If two files have the same data block, the data block from the first file to be backed up will be stored in the database.  The other file will just reference it.

    Note that when you run multiple dedup jobs, there is high demand on CPU and RAM.

  • Remember, the benefit to Backup to disk with BackupExec is that you can send multiple concurrent jobs to it.  Leverage that!  

  • .... provided you have the horsepower to handle the simultaneous jobs.  People often forget that there is no such thing as a free lunch.

  • well yer im seeing a bit of a slow down in performance with multiple dedupe jobs compared to multiple  backup-to-disk backups.

    server is an ibm x3650 32mb ram, intel xeon 3ghz, media server deduplication

  • Definitely, there is a lot more processing going before the data is stored in the dedup folder.  BE has to hash the data and check each block against existing blocks to make sure that it is unique before storing it. For B2D folders, it is just a straight write.

    When you run simultaneous jobs, you have to balance the load on your machine with the time gain.  With a not so poweful machine, you might be better off running consecutive jobs.

  • thanks pkh..

    not sure if this is possible but can a job automatically start when another is finished rather then having it based on a scheduled time in order to get around this issue?

    so really only one job for example has a start time and the rest just run consecutively after each has finished?