> I experienced that UNIONs in 7.1.1 are rather slow:
>
> tir=# explain (select nev from cikk) union (select
> tevekenyseg from log);
> NOTICE: QUERY PLAN:
>
> Unique (cost=667.63..687.18 rows=782 width=12)
> -> Sort (cost=667.63..667.63 rows=7817 width=12)
> -> Append (cost=0.00..162.17 rows=7817 width=12)
> -> Subquery Scan *SELECT* 1 (cost=0.00..28.16
> rows=1316 width=12)
> -> Seq Scan on cikk (cost=0.00..28.16
> rows=1316 width=12)
> -> Subquery Scan *SELECT* 2
> (cost=0.00..134.01 rows=6501 width=12)
> -> Seq Scan on log (cost=0.00..134.01
> rows=6501 width=12)
>
> Of course a simple SELECT is fast:
>
> tir=# explain select nev from cikk;
> NOTICE: QUERY PLAN:
>
> Seq Scan on cikk (cost=0.00..28.16 rows=1316 width=12)
>
>
> For me it seems to be slow due to the sorting. Is this right?
> Is this normal at all? Is it possible to make it faster?
If you know, that your result does not produce duplicates
(which are filtered away with "union") you can use a
"union all" which should be substantially faster, since it does
not need to sort.
Andreas