opcode

I tried it, and read striping is not the solution - as anticipated. DrivePool frequently reads from the slower, mechanical drive, even though all the data is also on an SSD. Would you consider adding a feature "For mirrored data, always read from drive X" (or, in more complex cases, allow to specify drive priorities, such that DrivePool will always read from the drive with the highest priority among the ones that contain the requested file). Should be reasonably simple to implement, shouldn't it?

Any chance we might see such settings in the future?

I don't think using read striping will result in optimal behaviour, because both a slow hard drive and a fast SSD would be in the same priority category ("Standard drive"), and therefore DrivePool would apply read striping across the two drives the slow drive could be the one with "the least outstanding requests" yet it could still be slower to use the drive for a new incoming request. Would you consider adding a feature that allows the user to control that mirrored files should always be read from the highest-priority drive, with priority being specified by the user? This would be a great addition for mirroring pools that use mixed drive types.

bump.

While DrivePool has lots of options to control file placement, i.e. writes (using SSD Optimizer, for example), I did not find corresponding features to control reads - beyond the striping option that one can enable. I use DrivePool on two of my systems to establish redundancy, i.e. I use it to mirror data only. At the moment, all my drives are approximately equally slow, but if I was to replace one of my pooled drives with an SSD, would it be possible to ensure that reads are performed primarily (or even exclusively) using the fastest drive? Thanks!