At present we have a single veeam server running Windows 2016 & 6 x 6TB drives in Raid 5. This gives us a 30TB Repo which we current use with Dedupe. We run a weekly active full which comes to around 5.4TB & Backups are moved to tape each week using the GFS feature. I get around 80% savings from the DeDupe alone, which is crazy!
However, given that Dedupe has caused me headaches before on other servers with corruption & the recent 2016 Bug that could have caused us to lose the repro (Thankfully we weren't affected!), along side the slow speed when recovering from a Deduped volume - I'm looking at moving over to something more resiliant. I know the RAID 5 I have at present isn't really the best, as the loss of a single disk puts me at risk & rebuild times are over 24 hours.
As I understand it, A storage space setup with REFS allows for auto healing of corrupted data, but you need to set the disks up in a mirror or parity format to do this. Does this mean I unraid all the disks & create the "array" from within Windows instead? If so, is this a supported solution on a single server & will disk performance struggle? Or should I simply recreate the Array in Raid 50 & a single REFS formated volume, which loses the corruption autorepair?