Bradley Tate wrote:
I'm getting an interesting result when restoring a 10G database on a Linux box. We've very recently installed 1.5.2 Classic after running 1.5.0 SuperServer for years. When we ran gbak `-R` on SuperServer 1.5.0 it used to take arout 4 hours. A restore on Classic seems to be taking well under 1 hour (closer to half that). No other jobs are running, no users on the box. I'm gobsmacked at the improvement and have a few theories but does anyone know why this is so?
Answer from Pavel Cisar:
I suppose you have used local connection? Then no wonder it's faster with Classic, because when local connection is used, classic is loaded directly into address space of client process, thus there is no overhead in data transfer between address spaces of different processes. And restore means a lot of data to transfer.