Re: Few big or many small datafiles per tablespace?

Thanks for your comments so far.

For my question I want to disregard aspects like
- any sort of partitioning

• managebility / administrability
• backup and recovery
• IO paths / physical disks / number of spindels (although these aspects are of course worth to be discussed in different threads).

I want to set the focus just on performance. If I look at your comments I conclude that it is nearly the same (few big vs. many small). Maybe there's a bit of an advantage for fewer larger files due to checkpoint operations with the need to update every file header. But isn't it a mostly theoretical point of view? Is this time difference really measurable?

Ronny Received on Fri Jul 11 2008 - 08:44:36 CDT