Tom Lane escreveu:
=?ISO-8859-1?Q?Andr=E9_Volpato?= <andre.volpato@xxxxxxxxxxxxxxxxxxxxx> writes:
Explain output:
HashAggregate (cost=19826.23..19826.96 rows=73 width=160) (actual
time=11826.754..11826.754 rows=0 loops=1)
-> Subquery Scan b2 (cost=19167.71..19817.21 rows=722 width=160)
(actual time=11826.752..11826.752 rows=0 loops=1)
Filter: (bds_internacoes(200805, 200806, (b2.cod)::text,
'qtdI'::text, 'P'::bpchar) >= 1::numeric)
-> Limit (cost=19167.71..19248.89 rows=2165 width=48) (actual
time=415.157..621.043 rows=28923 loops=1)
So I guess the question is "what is the bds_internacoes function, and
why is it so slow?"
This function is quite fast:
Aggregate (cost=5.17..5.18 rows=1 width=12) (actual time=0.286..0.287
rows=1 loops=1)
-> Index Scan using iinternacoes4 on internacoes (cost=0.01..5.16
rows=1 width=12) (actual time=0.273..0.273 rows=0 loops=1)
Index Cond: ((((ano * 100) + mes) >= 200801) AND (((ano * 100)
+ mes) <= 200806) AND ((cod_benef)::text = '0005375200'::text))
Filter: (tipo_internacao = 'P'::bpchar)
Total runtime: 0.343 ms
The problem is that its fired up against 29K rows, wich takes the
total runtime about 10s.
We are guessing that a dual core 3.0GHz will beat up a quad core 2.2,
at least in this environmnent with less than 4 concurrent queryes.
--
[]´s, ACV
