Обсуждение: Slow first query
I'm running postgresql 8.0 under WindowsXP for a development box (the production server isn't up yet, and will be linux, so I don't know that the problem will travel)... Using a PHP connection to it for a simple web app, if I leave the app alone for a bit (about 20 seconds I'd say) and do something that requires a query the next page load is slow, not 20 seconds slow, but 3-5 seconds.... but as long as I stay active the page reloads are instantaneous. Is there anything I can do about this, or should I just wait to see how it performs on my production server before I worry too much?
Rob Brenart wrote: > I'm running postgresql 8.0 under WindowsXP for a development box (the > production server isn't up yet, and will be linux, so I don't know that > the problem will travel)... > > Using a PHP connection to it for a simple web app, if I leave the app > alone for a bit (about 20 seconds I'd say) and do something that > requires a query the next page load is slow, not 20 seconds slow, but > 3-5 seconds.... but as long as I stay active the page reloads are > instantaneous. > > Is there anything I can do about this, or should I just wait to see how > it performs on my production server before I worry too much? Since you don't say what's causing it, difficult to say what the problem is. Might it just be caching, does your development box have a lot of memory free to cache disk blocks? Otherwise, it could be almost anything. You'll have to run some timing/traces on the various elements of your setup (php/webserver/pg). -- Richard Huxton Archonet Ltd
Richard Huxton wrote: > Rob Brenart wrote: > >> I'm running postgresql 8.0 under WindowsXP for a development box (the >> production server isn't up yet, and will be linux, so I don't know >> that the problem will travel)... >> >> Using a PHP connection to it for a simple web app, if I leave the app >> alone for a bit (about 20 seconds I'd say) and do something that >> requires a query the next page load is slow, not 20 seconds slow, but >> 3-5 seconds.... but as long as I stay active the page reloads are >> instantaneous. >> >> Is there anything I can do about this, or should I just wait to see >> how it performs on my production server before I worry too much? > > > Since you don't say what's causing it, difficult to say what the > problem is. Might it just be caching, does your development box have a > lot of memory free to cache disk blocks? > > Otherwise, it could be almost anything. You'll have to run some > timing/traces on the various elements of your setup (php/webserver/pg). > Yeah, I didn't say what's causing it because as of yet I haven't had time to do any serious analysis on it... I thought I'd throw it out there and see if anyone else had experience with it first. My guess is it's exactly what you're saying, some component or another is getting tossed from RAM to disk and slowing things down.
On Tue, 2005-07-19 at 16:31, Rob Brenart wrote: > Richard Huxton wrote: > > > Rob Brenart wrote: > > > >> I'm running postgresql 8.0 under WindowsXP for a development box (the > >> production server isn't up yet, and will be linux, so I don't know > >> that the problem will travel)... > >> > >> Using a PHP connection to it for a simple web app, if I leave the app > >> alone for a bit (about 20 seconds I'd say) and do something that > >> requires a query the next page load is slow, not 20 seconds slow, but > >> 3-5 seconds.... but as long as I stay active the page reloads are > >> instantaneous. > >> > >> Is there anything I can do about this, or should I just wait to see > >> how it performs on my production server before I worry too much? > > > > > > Since you don't say what's causing it, difficult to say what the > > problem is. Might it just be caching, does your development box have a > > lot of memory free to cache disk blocks? > > > > Otherwise, it could be almost anything. You'll have to run some > > timing/traces on the various elements of your setup (php/webserver/pg). > > > Yeah, I didn't say what's causing it because as of yet I haven't had > time to do any serious analysis on it... I thought I'd throw it out > there and see if anyone else had experience with it first. > > My guess is it's exactly what you're saying, some component or another > is getting tossed from RAM to disk and slowing things down. Also note that there are settings in Windows workstation that favor higher performance for the interactive graphical environment over the performance of server type processes. So, if you've got a lot of memory being used by your GUI and desktop apps, this is completely normal behaviour in the workstation version of Windows, and the fix is to deploy to a server version.