Cache location - best practices?

[_m.a-x]

Hi all!

I am using Duplicacy to back plain data files of various types and size. The cache default location is a mirrored pool of 2 NVMe SSDs.

I am concerned that the transient nature of cache data will wear the drives prematurely. I use my high speed SSD pool for VMs and other containers and I’d like to extend the life of the pool as much as possible.

What is the general consensus around cache data? Should I just use a dedicated “scratch” SSD for cache? What happens if the disk fails and gets replaced?

Thx!

[saspus]

There is very little wear. Data is written once, and never rewritten. Duplicacy chunks are immutable. Raid of SSDs is of course a huge overkill for this.

Put it on HDD if you still worry about wear. It is not performance critical.

Cache can be deleted any time. Literally any time. Pieces will be fetched from the storage as needed.

[sevimo]

In the current baseline implementation cache performance is not critical indeed. If you use my large data optimization PRs (or if these get incorporated into baseline) then cache performance will be important. But in either case, cache in is mostly WORM (write once/read many), so is well suited for SSDs.