Re: AW: [PEAR] DB class speed

Hi! i've check both connection time and only query time. But with non-persistant is pgsql very slow, query time take 1/5th of connection time, but it's still much much slower. In my aplication connection takes 0.0418 and query take 0.0150 seconds. But when i enable persistant everthing start to fly. If i execute query in pgsql console it takes only 0.34 mseconds, thats almost zero. So i understand time 3.4 mseconds wiht persistant from php, but i do not understand time 15 mseconds for same query when using non-persistant. So i think i have to use persistant but what happens next. Memory can be filed very quickly, don't you think. -- tia, Uros mailto:uros.gruber@sir-mag.com Thursday, October 10, 2002, 10:44:01 PM, you wrote: TL> Uros Gruber [mailto:uros.gruber@sir-mag.com] wrote: >> >> Hi! >> >> I'm making some testing over DB.php >> >> And i wan't to know if this is normal. >> >> I've made some script where I connect do SQL server >> and execute one simple (1row) query. >> >> I use pgsql and mysql. tables and indexes are same for both >> servers. So test table is identical >> >> Whe i execute those script i get some timing results >> >> PGSQL >> Total time: 0.0576 >> Execution time: 0.0185 >> >> MySQL >> Total time: 0.0086 >> Execution time: 0.0028 >> >> I want to know why is sucsh diference. Both servers runing >> with socket connections and using connnect (non-persistant) TL> I think the non-persistent DB connection is the cause of the difference. TL> Maybe pgsql connects take (that) much longer than mysql's ? TL> I think you should try to TL> a) benchmark only the queries and row retrieval (and/or get an extra bench TL> for connection creation) TL> b) do an application-level benchmark, that is with some real data and TL> production logic TL> Btw, are there some benches for DB vs PHP's builtin clients (MySQL, TL> Postgres, Informix), perhaps with PHP-Acellerator? TL> Thomas