I use WD Reds in my Synology NAS in a SHR configuration for daily backups. Should I be worried? Should I replace the disks with other models? What are the risks if I keep the current configuration?
Unfortunately, the nature of RAID means that you can be dealing with a lot of random writes. Further, when a drive fails and you want to replace it, you are pushing massive amounts of writes to the new disk.
A lot of this is due to the Shingling. Like shingles on your roof, the physical locations where data is written overlap. The problem is that if you need to update an already written location, you may also need to update surrounding bits. This is where the random write performance degradation comes from.
With my limited understanding, the new models will fail under intensive work-loads. For example adding them to existing RAID setups to replace a failed drive, where in the beginning they have to be sync-ed and they fail under load after a few hours (either by dropping in performance or not being recognized at all).
They will work reasonable well under lower loads (more common for home NAS setups).
The biggest problem, and what the article is trying to point out, is that Western Digital (and Seagate) are doing all they can to hide this info from customers. They even advertise the HDD as being NAS/RAID friendly, when they are clearly not meant for that type of loads.
However, they have cut corners / optimized costs that assumes the drive is not under constant pressure, but rather mostly idle. What this implies is that they likely just have a hybrid magnetic/ssd backend, which is what the outrage is about.
No need for hybrids. DM-SMR drives usually have a non-SMR zone acting as a write buffer, similar to the SLC write buffers in non-SLC SSDs.
https://blocksandfiles.com/2020/04/15/shingled-drives-have-n...