remote datastore vs sync jobs

[mokaz]

Hi there all,

This post is about trying to understand what would be faster and most efficient in terms of PBS network resources usages.

1st scenario:
- running a "site specific" PBS server with a mounted "datastore" over NFS located @HQ (tested, working, slow due to own network limitation)

2nd scenario:
- running a "site specific" PBS server with a mounted "datastore" over NFS located @SITE
- syncing that remote "datastore" as a sync job toward the @HQ main datastore (currently testing this...)

Let me know,
Kind regards & happy year end times!
m.

 

[Dunuin]

You want local (SSD) storage for PBS. Otherwise GC and re-verify performance will be terrible and stuff like Live Restore won't be fast enough to be usable. So better to have a dedicated PBS at each location and then work with sync jobs.

 

[mokaz]

You want local (SSD) storage for PBS. Otherwise GC and re-verify performance will be terrible and stuff like Live Restore won't be fast enough to be usable. So better to have a dedicated PBS at each location and then work with sync jobs.

Hi there, yes forgot maybe to highlight that this setup is held for lab's purposes hence, I can live with slow backup/restore and skip some features etc.. I'll take your advise about a PBS at each locations + sync jobs... Towards today's testing's, it seemed to me that the average transfers over sync jobs was better (double) than backing up on a remote NFS datastore (that over the same underlay/overlay network setup).

Thanks !