[postgis-devel] RE: join selectivity

strk at refractions.net strk at refractions.net
Tue Dec 14 02:22:44 PST 2004


On Tue, Dec 14, 2004 at 10:15:14AM -0000, Mark Cave-Ayland wrote:
> Hi strk,
> 
> Could you post the explain outputs for those queries too? I don't see why
> with an index scan consideration between two columns that the planner has to
> invoke four RESTRICT functions?

strk=# explain analyze select * from test1, test2 where test1.geom && test2.geom;
NOTICE:  LWGEOM_gist_joinsel called (returning 0.000005)
                                                  QUERY PLAN                    
--------------------------------------------------------------------------------------------------------------
 Nested Loop  (cost=3.27..105.84 rows=1 width=64) (actual time=0.217..39.305 rows=2700 loops=1)
   Join Filter: ("inner".geom && "outer".geom)
   ->  Seq Scan on test2  (cost=0.00..28.32 rows=132 width=32) (actual time=0.081..1.111 rows=108 loops=1)
   ->  Materialize  (cost=3.27..3.52 rows=25 width=32) (actual time=0.001..0.011 rows=25 loops=108)
         ->  Seq Scan on test1  (cost=0.00..3.25 rows=25 width=32) (actual time=0.043..0.129 rows=25 loops=1)
 Total runtime: 40.471 ms
(6 rows)

strk=# create index test2_gist on test2 using gist (geom gist_geometry_ops);    CREATE INDEX
strk=# explain analyze select * from test1, test2 where test1.geom && test2.geom;
NOTICE:  LWGEOM_gist_joinsel called (returning 0.000005)
NOTICE:  LWGEOM_gist_sel called
NOTICE:   no constant arguments - returning default selectivity
NOTICE:  LWGEOM_gist_sel called
NOTICE:   no constant arguments - returning default selectivity
                                                  QUERY PLAN                    
--------------------------------------------------------------------------------------------------------------
 Nested Loop  (cost=3.27..92.11 rows=1 width=64) (actual time=0.046..39.219 rows=2700 loops=1)
   Join Filter: ("inner".geom && "outer".geom)
   ->  Seq Scan on test2  (cost=0.00..28.08 rows=108 width=32) (actual time=0.009..0.198 rows=108 loops=1)
   ->  Materialize  (cost=3.27..3.52 rows=25 width=32) (actual time=0.000..0.013 rows=25 loops=108)
         ->  Seq Scan on test1  (cost=0.00..3.25 rows=25 width=32) (actual time=0.002..0.052 rows=25 loops=1)
 Total runtime: 40.307 ms
(6 rows)

strk=# create index test1_gist on test1 using gist (geom gist_geometry_ops);    CREATE INDEX
strk=# explain analyze select * from test1, test2 where test1.geom && test2.geom;
NOTICE:  LWGEOM_gist_joinsel called (returning 0.000005)
NOTICE:  LWGEOM_gist_sel called
NOTICE:   no constant arguments - returning default selectivity
NOTICE:  LWGEOM_gist_sel called
NOTICE:   no constant arguments - returning default selectivity
NOTICE:  LWGEOM_gist_sel called
NOTICE:   no constant arguments - returning default selectivity
NOTICE:  LWGEOM_gist_sel called
NOTICE:   no constant arguments - returning default selectivity
                                                  QUERY PLAN                    
--------------------------------------------------------------------------------------------------------------
 Nested Loop  (cost=3.27..92.11 rows=1 width=64) (actual time=0.052..38.867 rows=2700 loops=1)
   Join Filter: ("inner".geom && "outer".geom)
   ->  Seq Scan on test2  (cost=0.00..28.08 rows=108 width=32) (actual time=0.012..0.181 rows=108 loops=1)
   ->  Materialize  (cost=3.27..3.52 rows=25 width=32) (actual time=0.000..0.010 rows=25 loops=108)
         ->  Seq Scan on test1  (cost=0.00..3.25 rows=25 width=32) (actual time=0.002..0.032 rows=25 loops=1)
 Total runtime: 40.027 ms
(6 rows)


... MM.. my database might be corrupted somehow:

strk=# vacuum full analyze test1 ;
ERROR:  index "pg_toast_1181604_index" is not a btree

strk=# select version();
                                                version                         
--------------------------------------------------------------------------------------------------------
 PostgreSQL 8.0.0rc1 on i686-pc-linux-gnu, compiled by GCC gcc (GCC) 3.3.2 20031005 (Debian prerelease)
(1 row)


.. But I'm sure I used initdb !

--strk;
  

> 
> 
> Kind regards,
> 
> Mark.
> 
> ------------------------
> WebBased Ltd
> South West Technology Centre
> Tamar Science Park
> Plymouth
> PL6 8BT 
> 
> T: +44 (0)1752 791021
> F: +44 (0)1752 791023
> W: http://www.webbased.co.uk
>  
> 
> > -----Original Message-----
> > From: strk at refractions.net [mailto:strk at refractions.net] 
> > Sent: 13 December 2004 17:13
> > To: Mark Cave-Ayland
> > Cc: postgis-devel at postgis.refractions.net; 
> > pgsql-hackers at postgresql.org
> > Subject: Re: [postgis-devel] RE: join selectivity
> > 
> > 
> > On Mon, Dec 13, 2004 at 03:04:01PM -0000, Mark Cave-Ayland wrote:
> > > Hi strk,
> > > 
> > > > -----Original Message-----
> > > > From: strk at refractions.net [mailto:strk at refractions.net]
> > > > Sent: 13 December 2004 14:05
> > > > To: Mark Cave-Ayland
> > > > Cc: postgis-devel at postgis.refractions.net
> > > > Subject: Re: [postgis-devel] RE: join selectivity
> > > > 
> > > > 
> > > > On Mon, Dec 13, 2004 at 12:16:15PM -0000, Mark Cave-Ayland wrote:
> > > > > Hi strk,
> > > > > 
> > > > > (cut)
> > > > > 
> > > > > > > Taking a look at join selectivity...
> > > > > > > For a query like this:
> > > > > > >
> > > > > > >	SELECT id FROM table1, table2 
> > > > > > >	WHERE table1.geom && table2.geom;
> > > > > > >
> > > > > > > RESTRICT selectivity is invoked twice and
> > > > > > > JOIN selectivity is invoked once.
> > > > > > > The RESTRICT code is not able to find a costant 
> > part and thus 
> > > > > > > returns the default value (0.000005), JOIN 
> > selectivity so far 
> > > > > > > returns an hard-wired 0.1.
> > > > > > >
> > > > > > > Questions:
> > > > > > > (1) What should RESTRICT selectivity do in this case ?!
> > > > > 
> > > > > > Maybe that's how the planner decide what to do:
> > > > > > 	1) sequencially scan table1 and use index for each row
> > > > > > (RESTRICT)
> > > > > > 	2) sequencially scan table2 and use index for each row
> > > > > > (RESTRICT)
> > > > > > 	3) ... some other magic I'm missing .. (JOIN)
> > > > > 
> > > > > Indeed, you could be on the right lines here in thinking
> > > > the planner
> > > > > considers some form of individual scan on each first before
> > > > finalising
> > > > > on a plan type (although unless the tables are small I 
> > would have
> > > > > thought this would not have been an option). Does this 
> > > > change if you
> > > > > do a SET ENABLE_SEQSCAN = 'f' before the query?
> > > > 
> > > > Bingo.
> > > > Both ENABLE_SEQSCAN = 'f' or unavailability of an index make
> > > > the selectivity estimator calls go away. The join selectivity 
> > > > is called nonetheless (also in absence of indexes).
> > > 
> > > Right. So what you're saying is that if there is *no* GiST index on 
> > > *one* of the geom columns, or sequential scans are 
> > disabled, then the 
> > > calls to RESTRICT go away?
> > 
> > Index on a single table makes 2 calls to RESTRICT:
> > 
> >   strk=# select * from test1, test2 where test1.geom && test2.geom;
> >   NOTICE:  LWGEOM_gist_joinsel called (returning 0.000005)
> >   NOTICE:  LWGEOM_gist_sel called
> >   NOTICE:   no constant arguments - returning default selectivity
> >   NOTICE:  LWGEOM_gist_sel called
> >   NOTICE:   no constant arguments - returning default selectivity
> >   
> > Index on a both table makes 4 calls to RESTRICT:
> > 
> >   strk=# select * from test1, test2 where test1.geom && test2.geom;
> >   NOTICE:  LWGEOM_gist_joinsel called (returning 0.000005)
> >   NOTICE:  LWGEOM_gist_sel called
> >   NOTICE:   no constant arguments - returning default selectivity
> >   NOTICE:  LWGEOM_gist_sel called
> >   NOTICE:   no constant arguments - returning default selectivity
> >   NOTICE:  LWGEOM_gist_sel called
> >   NOTICE:   no constant arguments - returning default selectivity
> >   NOTICE:  LWGEOM_gist_sel called
> >   NOTICE:   no constant arguments - returning default selectivity
> >   
> > No index makes no calls to RESTRICT:
> > 
> >   strk=# select * from test1, test2 where test1.geom && test2.geom;
> >   NOTICE:  LWGEOM_gist_joinsel called (returning 0.000005)
> > 
> > > 
> > > > > It just seems strange for a <column> <operator> 
> > <column> clause to
> > > > > call a function involving a constant. Again, I'd 
> > probably ask on 
> > > > > pgsql-hackers just to clarify - I think Tom Lane was 
> > > > involved with the
> > > > > planner, so will be able to answer this one fairly quickly.
> > > 
> > > CCd to pgsql-hackers for clarification. BTW which version are you 
> > > developing against - 7.4 or 8.0?
> > 
> > 8.0.0RC1
> > 
> > --strk;
> > 
> > > 
> > > 
> > > Kind regards,
> > > 
> > > Mark.
> > > 
> > > ------------------------
> > > WebBased Ltd
> > > South West Technology Centre
> > > Tamar Science Park
> > > Plymouth
> > > PL6 8BT
> > > 
> > > T: +44 (0)1752 791021
> > > F: +44 (0)1752 791023
> > > W: http://www.webbased.co.uk
> > > 
> > 
> 
> 
> _______________________________________________
> postgis-devel mailing list
> postgis-devel at postgis.refractions.net
> http://postgis.refractions.net/mailman/listinfo/postgis-devel



More information about the postgis-devel mailing list