On Fri, Dec 15, 2006 at 12:15:59PM +0000, Gregory Stark wrote:
> There are various attempts at providing better timing infrastructure at low
> overhead but I'm not sure what's out there currently. I expect to do this what
> we'll have to do is invent a pg_* abstraction that has various implementations
> on different architectures. On Solaris it can use DTrace internally, on Linux
> it might have something else (or more likely several different options
> depending on the age and config options of the kernel).
I think we need to move to a sampling approach. setitimer is good,
except it doesn't tell you if signals have been lost. Given they are
most likely to be lost during high disk I/O, they're actually
significant. I'm trying to think of a way around that. Then you don't
need a cheap gettimeofday at all...
Have a nice day,
--
Martijn van Oosterhout <kleptog@svana.org> http://svana.org/kleptog/
> From each according to his ability. To each according to his ability to litigate.