Re: Why query plan is different? - Mailing list pgsql-performance

<div class="moz-cite-prefix">On 10.10.2016 17:31, Andrzej Zawadzki wrote:<br /></div><blockquote
cite="mid:bb1d234e-62ee-70e8-44ef-9b0919ec1268@wp.pl"type="cite"> Hi,<br /> Today, I noticed strange situation:<br
/><br/> The same query run on different servers has very different plan:<br /><br /> Q: SELECT b.* FROM kredytob b 
WHEREpesel = '22222222222'  ORDER BY b.id DESC LIMIT 1 <br /><br /> Slow plan:<br /><br /> "Limit  (cost=0.43..28712.33
rows=1width=4) (actual time=2574.041..2574.044 rows=1 loops=1)"<br /> "  Output: id"<br /> "  Buffers: shared
hit=316132read=110001"<br /> "  ->  Index Scan Backward using kredytob_pkey on public.kredytob b 
(cost=0.43..3244444.80rows=113 width=4) (actual time=2574.034..2574.034 rows=1 loops=1)"<br /> "        Output: id"<br
/>"        Filter: (b.pesel = '22222222222'::bpchar)"<br /> "        Rows Removed by Filter: 433609"<br /> "       
Buffers:shared hit=316132 read=110001"<br /> "Planning time: 0.414 ms"<br /> "Execution time: 2574.139 ms"<br /><br
/><br/> Fast plan:<br /> "Limit  (cost=115240.66..115240.66 rows=1 width=4) (actual time=463.275..463.276 rows=1
loops=1)"<br/> "  Output: id"<br /> "  Buffers: shared hit=14661 read=4576"<br /> "  ->  Sort 
(cost=115240.66..115240.94rows=112 width=4) (actual time=463.271..463.271 rows=1 loops=1)"<br /> "        Output:
id"<br/> "        Sort Key: b.id DESC"<br /> "        Sort Method: top-N heapsort  Memory: 25kB"<br /> "       
Buffers:shared hit=14661 read=4576"<br /> "        ->  Index Scan using kredytob_pesel_typkred_opclass_idx on
public.kredytobb  (cost=0.43..115240.10 rows=112 width=4) (actual time=311.347..463.183 rows=5 loops=1)"<br />
"             Output: id"<br /> "              Index Cond: (b.pesel = '22222222222'::bpchar)"<br /> "             
Buffers:shared hit=14661 read=4576"<br /> "Planning time: 0.383 ms"<br /> "Execution time: 463.324 ms"<br /><br /> Data
isalmost equal - "slow" has a few more rows in table. ("Fast" is a copy from 1 am today).<br /> Why runtime is
slower?<br/></blockquote><br /> I made another INDEX, without opclass:<br /><br /> CREATE INDEX
kredytob_pesel_typkred_idx<br/>   ON public.kredytob<br />   USING btree<br />   (pesel COLLATE pg_catalog."default",
typkred);<br/><br /> after that: analyze kredytob;<br /><br /> And now:<br /> "Limit  (cost=333.31..333.31 rows=1
width=4)(actual time=0.100..0.102 rows=1 loops=1)"<br /> "  Output: id"<br /> "  Buffers: shared hit=8"<br /> "  -> 
Sort (cost=333.31..333.59 rows=114 width=4) (actual time=0.095..0.095 rows=1 loops=1)"<br /> "        Output: id"<br />
"       Sort Key: b.id DESC"<br /> "        Sort Method: top-N heapsort  Memory: 25kB"<br /> "        Buffers: shared
hit=8"<br/> "        ->  Index Scan using kredytob_pesel_typkred_idx on public.kredytob b  (cost=0.43..332.74
rows=114width=4) (actual time=0.046..0.065 rows=5 loops=1)"<br /> "              Output: id"<br /> "              Index
Cond:(b.pesel = '22222222222'::bpchar)"<br /> "              Buffers: shared hit=8"<br /> "Planning time: 0.438 ms"<br
/>"Execution time: 0.154 ms"<br /><br /> So, what is a reason that "SLOW" server doesn't like opclass index?<br /><br
/>-- <br /> Andrzej<br />