Subject | Re: Can we, can we, can we????... |
---|---|
Author | johnson_dave2003 |
Post date | 2005-06-21T01:27:54Z |
--- In Firebird-Architect@yahoogroups.com, David Johnson
<d_johnson@c...> wrote:
Currently, we limit on CPU (the operating system does measure true CPU
- it obviously isn't a WIntel).
I am not sufficiently familiar with Firebird internals yet to have a
solid suggestion. Tentatively, I would suggest count of passes
through the main loop. Alternatively, chronological time spent in the
main loop, as measured with the high resolution timer (where
available). I have a DLL around here for capturing high resolution
timer on Intel boxes. Its overhead is measured at 0.002 milliseconds
per call on a 2 GHz Dell desktop. I have also seen something like it
compiled as a .so for linux.
<d_johnson@c...> wrote:
>...
> > Ann wrote:
> >
> Perhaps I'm being obtuse, but I don't see exactly where quotas wouldI forgot to answer your question ...
> help. The ability for a developer to kill a query - absolutely, but
> what quota would you choose to limit?
>
Currently, we limit on CPU (the operating system does measure true CPU
- it obviously isn't a WIntel).
I am not sufficiently familiar with Firebird internals yet to have a
solid suggestion. Tentatively, I would suggest count of passes
through the main loop. Alternatively, chronological time spent in the
main loop, as measured with the high resolution timer (where
available). I have a DLL around here for capturing high resolution
timer on Intel boxes. Its overhead is measured at 0.002 milliseconds
per call on a 2 GHz Dell desktop. I have also seen something like it
compiled as a .so for linux.