Capacity tier stuff

[backupquestions]

1. If I take monthlies of a file server locally on a ReFS file system drive, taking barely more than the space of one full... What is the need for cloud tier? Would each reported (but not used due to refs) file size of each full vbk still be sent to the cloud tier without a problem once it ages out? To me, it almost makes more sense to keep all GFS local using ReFS since there isn't a space issue?? Then when you factor in that cloud tier is not a true copy like VCC is, its even more so... Could I get some explanation on that? Or why I would want to use it?

2. My plan right now is to use VCC for a 1 week retention on all VMs... Then I have a handful of file servers that have longer term requirements... So what would be the best route if I want redundancy in my long term backups? My only thoughts of how to do this are a second local repository or use cloud tier for all GFS stuff, but then all the GFS points are only in one true spot (cloud) and not in local.

What are some customers doing similar to my situation?

[Gostev]

1. Cloud Tier works the same way as ReFS, so disk consumption will be the same. Provided built-in redundancy with multiple copies of blocks and relatively low object storage cost some cloud providers are able to offer, this provides an attractive target for offloading older backups from generally more expensive on-prem storage.

2. If you don't feel having all GFS restore points in object storage is sufficient, you can always set up additional Backup Copy jobs to go to an on-prem storage, or do a tape outs. Most customer seem to prefer tape for these extra copies to achieve media break, instead of yet another copy on disk.

[backupquestions]

Thank you. So to clarify more on your first sentence...

If I have a local ReFS repo, with 6 monthlies of a 1TB system... Locally this will appear as 6 800GB VBK files or whatever size it ends up as etc... Then, with cloud tier the first offload is 800GB and then all the other GFS points are only tiny changes between them but still sealed full vbks? Rather than 6 separate 800GB files?

So overall as you mentioned in 2., having all gfs restore points existing only in object storage in the cloud is a great option vs paying whatever it would cost for the same retention in VCC? VCC is meant more for short term retention copy but can do GFS if needed etc. ?

[Gostev]

Yes and Yes.

VCC = Offsite backup (can use short-term or long-term retention).
Cloud Tier = Capacity extension for an on-prem repository.

[backupquestions]

Thanks again.

[poulpreben]

1. Cloud Tier works the same way as ReFS, so disk consumption will be the same. Provided built-in redundancy with multiple copies of blocks and relatively low object storage cost some cloud providers are able to offer, this provides an attractive target for offloading older backups from generally more expensive on-prem storage.

Does it also work the other way? I.e. if I retrieve all archived data from Cloud Tier, will it leverage fast clone on my ReFS file system? My use case is to temporarily use object storage for migrating from a deduplicated NTFS volume to ReFS.

[Gostev]

I believe it only works when you're "restoring" data back to the stub files on the original SOBR extent. We will double check.

[veremin]

I believe it only works when you're "restoring" data back to the stub files on the original SOBR extent. We will double check.

Unfortunately, it doesn't work today, but not because it's not supposed to, but because of some bug in current logic. We're planning to fix it in the next release, though. Thanks!

[backupquestions]

I have tested capacity tier and it works great. I was only able to test using active fulls for my sealed chains so far.

My question is...

1. If I have a backup copy job with gfs enabled and these gfs points are refs spaceless fulls.. these age out to capacity tier..... now... let's say I want to move them back to performance tier. When they re hydrate locally do they lose spaceless or do they go back to their spaceless full size? E.g. 1tb size but consuming 10g space etc.

I'm thinking they would never come back spaceless again due to block cloning loss but I guess I never really need them locally anyway. These are gfs points only out there for retention requirements and only file level restore would ever be needed.

[veremin]

It will work the way you've described, once we fix the bug mentioned above (should happen in one of the next product releases). Thanks!