[vox-tech] querying postgres from php

Dylan Beaudette dylan.beaudette at gmail.com
Tue Nov 14 17:07:06 PST 2006


On Tuesday 14 November 2006 12:45, Henry House wrote:
> På 2006-11-14, skrev Dylan Beaudette:
> > Hi everyone,
> >
> > i have a large postgres table, that i am querying from php: specifically:
> >
> > 1. query the rule classes associated with a given ID
> > 2. query for a set of rules within a given class
> > 3. for each rule, query the reasons why that rule is enforced
> >
> > this results in about 60 'select ...' statements. Watching the queries as
> > they are run from pgtop, i notice that they run serially and not all that
> > fast.
> >
> > Running the same set of queries from the psql command line client takes
> > about 5 seconds max. Therefore, I think that the problem may be in the
> > overhead involved in executing each query- or some PHP/Apache thing. I
> > have a full set of indices built on this table, and using 'explain query
> > ...' yields no 'seq scans', suggesting (i think) efficient use of the
> > indices.
> >
> > Any thoughts on how one might approach this issue ?
>
> I am assuming that you have already tried to combine the about 60
> queries into a smaller number of queries and that there is a good reason
> why this can't be done.
>
> My initial though is to try preparing the queries with placeholders for
> their specific parameters prior to execution. I don't know anything
> about database access in PHP, but in Perl and Ruby you can do something
> like this (pseudocode, not actually runnable):
>
> statement_handle = db_handle.prepare("select foo from bar where foo = ?")
> result1 = statement_handle.execute(statement_handle, "a")
> result2 = statement_handle.execute(statement_handle, "b")
> result3 = statement_handle.execute(statement_handle, "c")
>
> which is equivalent to:
>
> result1 = db_handle.execute("select foo from bar where foo = 'a'")
> result2 = db_handle.execute("select foo from bar where foo = 'b'")
> result3 = db_handle.execute("select foo from bar where foo = 'c'")
>
> The former case should be faster as it allows the database engine (if so
> designed) to re-use the same query plan for the same logistically
> equivalent queries.

Thanks henry!

Fortunately I was able to consolidate things down to just a couple queries, 
with logic in PHP to do the dirty work.

Cheers,

-- 
Dylan Beaudette
Soils and Biogeochemistry Graduate Group
University of California at Davis
530.754.7341


More information about the vox-tech mailing list