Open Dikay900 opened 1 year ago
You can do this by using foreground_target
to direct writes to your fast devices, background_target
to balance things to your slow devices, and then use promote_target
to cache data from your slow devices in your fast ones' free space.
First of all big thanks for your work on this fs.
In the principles of operation you state:
by default
sounds like you can change this behavior, so is this already possible to setup or rather something for the future?I would like to have the maximum of write speed for multiple devices which have different size and performance characteristics. So the fastest devices will fill fastest and then the data gets balanced, which would make manually setting up a promote_target unnecessary or rather optional.
My user scenario is having mixes of ssds and nvmes which are all having slightly different performance characteristics which will slow down the system to the slowest device for the promote_target. This is obviously the same as the allocator does this in the same way for all devices when using no promote_target.