Source and target replication - setting different retention policy

Running 6.3

I have a source and target configured at two separate locations with replication. I want to be able to split my recovery points to that my hourly, daily backups are stored on the source, but then my weekly, monthly, yearly are kept at the target. Having trouble wrapping my head around this...Is this how it would be done?
Source Policy:
Keep all recovery points for 3 days.
…and then keep one recovery point per hour for 3 days
…and then keep one recovery point per day for 4 days
…and then keep one recovery point per week for 1 weeks
…and then keep one recovery point per month for 0 months
…and then keep one recovery point per month for 0 year
Target Policy:
Keep all recovery points for 3 days.
…and then keep one recovery point per hour for 3 days
…and then keep one recovery point per day for 4 days
…and then keep one recovery point per week for 3 weeks
…and then keep one recovery point per month for 2 months
…and then keep one recovery point per month for 1 year

Thanks

Parents
  • I highly recommend not setting your source core retention policy to have 0 RPs per month or per year. The reason is that this means every week you will have all of the base images on your source core roll up into some incrementals. That means there will be a significant amount of data processing each week and it's likely that the maintenance jobs like rollup and deferred delete will fall behind. Instead I would set the "...and then keep one recovery point per year" to 1 year. That way you only rollup your base images once per year as opposed to every week. Also, the amount of data that causes you to keep additionally is small in comparison to the workload it would put on your system to have the retention policy you suggested.

    On the target core you could set the hourly and daily recovery points to 1 so that you aren't keeping lots of those. You stated you only want the weekly, monthly and yearly on the target, so you could uncheck everything except the ones you want to keep and the core will obey that policy. You cannot uncheck keep all recovery points because we have to have all the latest recovery points in order to do the merges necessary to honor the retention policy, but you can definitely uncheck the hourly and daily sections.

Reply
  • I highly recommend not setting your source core retention policy to have 0 RPs per month or per year. The reason is that this means every week you will have all of the base images on your source core roll up into some incrementals. That means there will be a significant amount of data processing each week and it's likely that the maintenance jobs like rollup and deferred delete will fall behind. Instead I would set the "...and then keep one recovery point per year" to 1 year. That way you only rollup your base images once per year as opposed to every week. Also, the amount of data that causes you to keep additionally is small in comparison to the workload it would put on your system to have the retention policy you suggested.

    On the target core you could set the hourly and daily recovery points to 1 so that you aren't keeping lots of those. You stated you only want the weekly, monthly and yearly on the target, so you could uncheck everything except the ones you want to keep and the core will obey that policy. You cannot uncheck keep all recovery points because we have to have all the latest recovery points in order to do the merges necessary to honor the retention policy, but you can definitely uncheck the hourly and daily sections.

Children
  • Great catch, I forgot about that. I know this is probably a fairly small end case that Monthly/ Yearly would be set to 0 but shouldn't there be some type of warning about this? The Core should not let you do something that is going to probably cause issues without some type of warning. If you set Monthly or Yearly to 0 a dialog pops up explaining the impact and work-arounds?

    Or better yet, put in some type of fix. Maybe an option to allow Monthly/ Yearly to be 0 but if you check "Yearly Rollup" it still only tries to rollup the bases once per year. No idea if that would work but just an example. 

  • You are more than welcome to put in a feature request for those ideas at https://ideas.labs.quest.com.

    It is very possible that rolling up the base image every week won't be a problem. It all depends on the resources that the OP has, how big the backups are, how frequently backups are happening, etc. As with everything, it's all about balancing the hardware resources with the software configuration.

  • Thanks all for the feedback - good stuff.