Hi all, I'm using a single AUFS cache_dir with 800GB on a RAID-0, 3 15K RPM disks. On peak hours it doesn't work very well, now that it has around 80GB of files (200,000 objects). The mean object size is 400KB. I've been reading on how I can improve performance on peak hours, and I'm considering spreading on a few cache_dirs with: store_dir_select_algorithm round-robin I'm thinking about doing around 10 80GB cache_dirs... Does anybody knows if it can help? Cheers, - Robert
