Question About AWS3 Retention

[cmanrique]

So we currently have a backup job that points to an SOBR which has a AWS3 bucket. This job has a retention policy of 28 restore points and it creates an active full backup monthly on the first Friday of each month, so it is in forward incremental backup mode.

The SOBR is configured as follows: it moves backups to object storage as they age out of the operational restore window, which makes it move backup files older than 28 days.

Currently, the job is working the way we want it, which is we want to have a month's worth of backups on premise locally and as they age out, they get sent to the AWS3 capacity tier. However, as I've noticed with another job we have, backups in AWS3 will start to get deleted once we have 28 retention points on AWS3 due to the retention policy. We do not want this to happen, since we want a year's worth of backups residing on AWS3.

My question then is, how can we keep the 28 restore point retention on premise but also have a year's worth of backups on AWS3? Another post had mentioned GFS retention, but I don't totally understand how it works, so if that is the solution, please let me know the best way to configure it so that our goal is met.

[Gostev]

1. Change the retention policy of your backup job to the desired one (I understand it is 1 year in your case).

Most people use GFS for long term retention (see "Keep certain full backups longer for archival purposes" checkbox), but this is optional of course... you can instead just set your retention policy in the backup job settings to 365 days or restore points.

2. Enable the Move policy in the Capacity Tier settings with the 28 days window.

[cmanrique]

Thank you, I will try the solution from this thread when I need the long term retention: post377251.html?hilit=retention#p377251

Otherwise, I will try your solution. I do have a question about it though. With it, each active monthly backup will continue to have a month's worth of incrementals correct? And the incrementals from the previous months will start being sent to the Capacity tier and leave behind locally a small metadata file, correct? In other words, locally, I can expect to have about a cumulative file size of about a month's worth of backups/incrementals and not more than that despite the previous month's incrementals?

[Gostev]

Actually, the solution from that thread is exactly the same as I suggested above

No, with monthly fulls you will have two months worth of backups on-prem in the worst case scenario. Due to how incremental backup works, we cannot remove the entire chain until the latest increment in the chain is still under the retention policy. The User Guide for the Capacity Tier has a good explanation of this too.

To avoid this issue, I recommend you use ReFS or XFS repository with weekly synthetic fulls, along with your monthly active full. Synthetic fulls on ReFS/XFS don't take any physical disk space, but their presence will allow you to inactivate previous backup chains without waiting for an additional month for the last increment in chain to expire.

[cmanrique]

Ok thanks, we can deal with the 2 months of backups. I will try GFS with the other Backup Job we have, which has 1 retention point and sends a copy of it to the Capacity tier right when it's created once a week. It has 1 retention point since we don't want it to have incrementals, just full backups, and we only want to have 1 locally. It doesn't work as we want it right now, since every week it deletes a backup from AWS3, and it only has 2 backups there. If I understand GFS correctly, if I set a flag for weekly backups and keep 52 weeks of backups, along with the 1 retention point and changing the option in the SOBR from "copying the backup to AWS3 as soon as it's made" to "moving backups older than a day old", then each backup should successfully stay in AWS3 for a year.

[Gostev]

You will want to keep the Copy policy on (to have a copy of your latest backup in the cloud), and enable the Move policy in addition to it (to remove non-current backups from on-prem storage).