Question ECC-capable hardware for DIY NAS ?

[Dreadwolf91]

So i read about this doomsday scenario where not having ECC-capable hardware might lead to corruption that could delete the whole storage pool especially with ZFS.

I wonder, when i have an external HDD where i automatically backup all my important files from the NAS weekly do i risk infecting this external hdd with the dangerous corruption?
Or is the data corruption only dangerous for the system it appeared on ?

As context, i wanna build myself a Server, mostly to use as a NAS but with a bit more decent hardware like an i3 or a ryzen 5500. i will have like 2tb (which is alread much more than actually necessary) for photography and documents and like 20tb for my self-ripped blurays. worst case i can re-rip my blurays but photography and documents need to be safe and i cant risk to "infect" every backup with a possible data corruption.

 

[CountMike]

So i read about this doomsday scenario where not having ECC-capable hardware might lead to corruption that could delete the whole storage pool especially with ZFS.

I wonder, when i have an external HDD where i automatically backup all my important files from the NAS weekly do i risk infecting this external hdd with the dangerous corruption?
Or is the data corruption only dangerous for the system it appeared on ?

As context, i wanna build myself a Server, mostly to use as a NAS but with a bit more decent hardware like an i3 or a ryzen 5500. i will have like 2tb (which is alread much more than actually necessary) for photography and documents and like 20tb for my self-ripped blurays. worst case i can re-rip my blurays but photography and documents need to be safe and i cant risk to "infect" every backup with a possible data corruption.

Consumer CPUs do not use ECC so nothing to worry about even if RAM is capable, also no reason to explicitly choose ECC capable RAM-

 

[Misgar]

If you try to copy corrupted files from your ZFS system to a NAS, you may find the copy process halts if checksums fail when reading one or more data blocks in damaged files. Whether this damage occurred due to memory errors or hard disk errors is immaterial. You've probably lost your file.

I have two old HP servers repurposed for TrueNAS Core RAID-Z2 arrays (one with 6 x6TB drives, the other with 8 x 4TB drives). The HP servers contain ECC RAM and Xeon CPUs. I have two other TrueNAS CORE RAID-Z2 arrays built using standard non-ECC desktop mobos and LSI HBA SAS controllers, each driving 8-disk arrays.

Although ECC RAM provides a slight improvement in data integrity under TrueNAS, there are many other things that can go wrong besides flipped bits in RAM, e.g. disk failure or controller failure.

In theory I can afford to lose two drives out of each RAID-Z2 array and all my data will still be intact, but in practice more errors can appear when resilvering the array after drive failure(s).

You should not save just one copy of important data to ZFS, then backup at a later date from the ZFS to NAS. Instead, make two or three copies direct from the original source to different media/systems. That way you're not copying from source to ZFS, then from ZFS to NAS, with the risk that corruption in ZFS will prevent subsequent backups to NAS. I backup to other PCs and 800GB LTO4 tapes in an external SAS tape drive.

Remember, RAID is not a backup if it's your only copy.