Raw vs. Cooked files
A user installing IDS 10 on Red Hat Enterprise Linux 3 (x86) asked whether to use raw devices or cooked files for dbspaces. He quoted an old FAQ entry (Art Kagel, 1998) showing raw disk benchmarks clearly faster than cooked/filesystem files because writes bypass the UNIX buffer cache, and asked whether that still held given improvements in filesystems had supposedly narrowed the gap. The thread contains only this one posting; no replies or resolution are recorded.
Auto-generated by DrWatson from the posts below — may be imperfect; read the full thread.