Bucket S3 compatible

[matteu]

Hello,

I will use S3 compatible (netapp) for backup copy with object lock30 days for my "vmware VMs" and pre installed agent Windows.

I have 2 different Veeam server because 2 isolated infrastructure but they both share the same S3 Netapp.
Bucket size is 30To / environment

I also have to backup Netapp filer to Hardened linux repository and second target the Netapp S3. I can read in the documentation 14 days is the recommended value for unstructured data when object lock is used (my scenario).
https://helpcenter.veeam.com/docs/backu ... ml?ver=120



Do I need to use the same bucket and the same folder for all my VM jobs + the unstructured and set the value to 14 (only 1 repository added) ?
Do I need to use the same bucket for both but 1 folder for VM / Agent + 1 folder for unstructured (2 repository 1 with 30 days and the other with 14 days) ?
Do I need to use 2 buckets 1 for VM / agent + 1 for unstructured (2 repository 1 with 30 days and the other with 14 days) ?

1 ) What would recommend in this scenario ?

2) Can you confirm me if full backup are needed or not when I use S3 for backup copy for VM ? because I can see if I don't check GFS, I can continue the wizzard without any warning. With Hardened linux repository, the wizzard prompt me to enable Full backup to have immutability.
When I read the documentation here, I'm not sure full are needed for object storage immutability to work : https://helpcenter.veeam.com/docs/backu ... ml?ver=120

Thanks

[david.domask]

Hi matteu,

1. Dedicated buckets is usually recommended depending on the size of the workload -- the performance can drop dramatically, and we have a limitation that buckets must not be shared (see the Important Note section on this page)

2. You can work with Forever Forward chains if you like, and it's even the Default that it is expected to work with, so it's your choice. Fulls are not required, but naturally GFS is recommended if there is a need for archival restore points.

[matteu]

Hello,

Thanks for your answer.

Can you élaborate more dedicated bucket please ?
1 for vm and an other for unstructured data ?
I don t use SOBR.
Each veeam backup server hold a dedicated bucket.
My question is to know for 1 vbr server do I need to use several bucket and if yes, how much for what use case.

Thanks !

[matteu]

For you to have all the détails :

Vbr1 backup vmware vm + 1 preinstalled windows agent
Backup from storage snapshot to HLR number 1 and backup copy to netapp S3 on bucket 1 and folder 1

Vbr2 backup vmware vm + unstructured data.
Backup from vmware snapshot to HLR number 2 and backup copy to netapp S3 on bucket 2 and folder 2

Is it correct or do I need to cfeate more bucket or separate folder ?

I also saw I need to use 1MB block size for storage object.

[david.domask]

Hi matteu,

Sure I can elaborate. In your original question, you asked about same bucket + same folder, though perhaps I misunderstood your intent, but the point is that a single bucket + folder should be used only for one repository; do not make multiple repositories pointing to the same bucket + folder.

Similarly, as per the User Guide, do not add the same bucket to multiple Veeam installations to use as a repository.

That's the main concern you need to avoid. In general, one bucket per repository makes sense, and ensure you're avoiding the considerations mentioned in the User Guide link I placed before.

From your post above this one, that looks correct and is in line with what I was advising, so I think I misunderstood your original description. But main points are as above.

[matteu]

Hello,

Thanks for your answer

Perfect, my design was correct 1 bucket / VBR ^^

Is there any advantage to create multiple folders inside the same bucket for 1 dedicated VBR?

[Gostev]

Depending on object storage, it may help scalability as some object storage chunks metadata catalogs based on prefixes (afaik AWS S3 does this). A more universal approach to scalability however is creating separate buckets, this should help with literally any object storage.

[matteu]

Thanks for your answer