Possibility to do file-level deduplication

[Adrigrou]

Hi PBS community,

I am needing some advices please.

Is there a way to do backup on file level without chunking ?
Or to use the file as the chunk ?

The backups will be done with proxmox-backup-client on physical servers.

Or maybe, what could be the ideal set-up for my backup servers?

2*xeon 6230
24*16TB SAS HDD
2*400Gb SAS SSD ( system )
2*1TB SAS SSD ( Special device )
not sure : 512GB-2TB of RAM

6 servers like that; each one mirror => 3 master backup servers and 3 replicates servers

And i am almost sure that i am gonna have some issue with chunk evaluation and creation when doing backup... but i may be wrong. I haven't tested PBS on anything other than full SSD configuration for the moment.

Does anyone have this kind of set-up ? does it work fine ? what rule should i follow, for number of disk/ size / number of special disks / size / ram size

I am not gonna use NVME or SSD for the main storage ( only for the special dev )

Thank you

 

[ggoller]

Hi!

And i am almost sure that i am gonna have some issue with chunk evaluation and creation when doing backup... but i may be wrong. I haven't tested PBS on anything other than full SSD configuration for the moment.

What you makes you think this?
A backup on file level without chunking is not possible. You could have a look at these section though: https://pbs.proxmox.com/docs/technical-overview.html#dynamically-sized-chunks and
https://pbs.proxmox.com/docs/backup-client.html#change-detection-mode. It explains how we select the chunk boundaries and how you can select to include files based on if their metadata changed (so files which metadata didn't change won't be touched).

Obviously we recommend SSD's but if you have decent HDD's (+ special device SSD's) it should work quite well.