Hello Heitor, thanks for your quick response - I have seen your videos on YT!
> On 13. Mar 2022, at 23:44, Heitor Faria <hei...@bacula.com.br> wrote: > >> Hello fellow users, > > Hello Justin, > >> I will only use disk storage and S3 buckets, I do not own any tape drives. >> >> There are up to 10 FDs (Linux and macOS), one of the SDs runs on the same >> machine as the Director and the SD which is going to write to the disks and >> to >> the S3 bucket(s). The remote S3 bucket will also be implemented by myself >> using >> minIO. > > Bacula can work with minIO, but you don't need. Bacula S3 Driver can write to > any S3 compatible cloud. Certainly. However, I do want my data under my own authority - only. > >> The Linux FD running on the same host as the Dir and SD is going to have jobs >> for: >> -onsite SMB shares for Synology servers in a DMZ, so that I do not need to >> run >> an FD in the DMZ which would need to reach into my LAN. >> -diverse non-Bacula backups form diverse systems, such as ESXi ghettoVCB, >> macOS >> TimeMachine, onsite minIO S3 buckets used by Synology DSM HyperBackup > > GhetoVCB is a very well-intentioned, but extremely poor solution to perform > VSphere VM backups in terms of performance and features. If you want a 100% > open source solution, I would migrate to Proxmox. Sure, I know about the Proxmox backup features. Still, for the next time I will run on ESXi, and this need a solution for that platform. >> Each backup job is intended to create 3 redundant “copies” (not meaning the >> Copy >> concept of Bacula) for safety reasons: >> - Tier 1 (fastest): 1 to n internal SATA hard disk drives (ideally no manual >> intervention when individual disks 1,.., n-1 fill up) >> - Tier 2 (medium speed): 1 to m external USB hard disk drives (ideally no >> manual intervention when individual disks 1,.., n-1 fill up) >> - Tier 3 (slowest): 1 or more offsite S3 buckets implemented using minIO >> Docker >> container on Synology DSM > > I would use one Cloud Storage with the CacheRetention directive for tiers 1 > and 3. External USB (ew!) can be a Bacula Copy Job or a completely different > redundant Job with its own Full and partials chain (e.g. Job1-cloud; > Job1-usb). Let’s see if I understand that correctly. What you are saying is that I should treat tier 1 as the cache for tier 3 and set the retention periods properly. Right? Yes, USB is terrible, yet a cheap solution for me, as I have some seriously sized HDDs laying around which were retired from some arrays due to size reasons, not for error reasons. I have put them in external cases with USB 3.1 Gen2 controllers and connected them using a USB 3.0 hub. As long as the jobs do not write to several HDDs simultaneously, performance wouldn’t be too bad. What are the drawbacks of a Copy job as compared to a independent Backup job? If I understand correctly that independent Backup jobs would tie up client and network resources, and Copy jobs do not. What’s the catch with the Copy jpb results to be used as tier 2? >> As I have the luxury of more than 1 drive per tier, I would like to >> understand, >> if it is possible to define a pool on top of more than one disk device (e.g. >> using AutoChanger with disk devices instead of tape devices?), so that first >> one device is used and if it is full, further volumes are automatically >> written >> to the next disk device (without the need for manual intervention). > > As for disks, I would just use LVM or any sort of RAID to present Bacula. > More practical. Brilliant, will do that! (Sometimes I get drawn into a rabbit hole of complexity while overlooking an easier and better solution.) Thank you for your time and support! JC _______________________________________________ Bacula-users mailing list Bacula-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/bacula-users
