Обсуждение: Efficiency again...

Поиск
Список
Период
Сортировка

Efficiency again...

От
Michael Richards
Дата:
Hi.
I just noticed something interesting. I don't know if my idea is better or
if it wasn't implemented because it violates some SQL rule...

searchengine=> create table test ( test1 int4, test2 int4);
CREATE
searchengine=> create index test_itest1 on test (test1);
CREATE
<insert a pile of data so it looks like so>
searchengine=> select * from test;
test1|test2
-----+-----
    1|    3
    1|    5
    1|    9
    2|    1
    2|    3
    2|    6
    2|    9
    3|    9
    4|    5
(9 rows)

Now here is the plan I expect for a single test1 value
searchengine=> explain select * from test where test1=1;
Index Scan on test  (cost=0.00 size=0 width=8)

But look:
searchengine=> explain select * from test where test1=1 or test1=2;
Seq Scan on test  (cost=0.00 size=0 width=8)

ugh! Sequential. This may be OK for a small database, but in my
application I have many rows:
searchengine=> explain select * from word_detail where word_id=23423 or
word_id=68548;

Seq Scan on word_detail  (cost=205938.73 size=510342 width=10)

That costs a _LOT_.

Wouldn't it be better to do n sequential scans where n is the number of
or'd together values? Using IN doesn't help out either...

searchengine=> explain select * from test where test1 IN (5,9);
Seq Scan on test  (cost=0.00 size=0 width=8)

Sometimes I wish I had the power to tell the DBMS how I wanted a query
done...

-Mike


Re: [HACKERS] Efficiency again...

От
Bruce Momjian
Дата:
> Now here is the plan I expect for a single test1 value
> searchengine=> explain select * from test where test1=1;
> Index Scan on test  (cost=0.00 size=0 width=8)
>
> But look:
> searchengine=> explain select * from test where test1=1 or test1=2;
> Seq Scan on test  (cost=0.00 size=0 width=8)
>
> ugh! Sequential. This may be OK for a small database, but in my
> application I have many rows:
> searchengine=> explain select * from word_detail where word_id=23423 or
> word_id=68548;
>
> Seq Scan on word_detail  (cost=205938.73 size=510342 width=10)
>
> That costs a _LOT_.
>
> Wouldn't it be better to do n sequential scans where n is the number of
> or'd together values? Using IN doesn't help out either...
>
> searchengine=> explain select * from test where test1 IN (5,9);
> Seq Scan on test  (cost=0.00 size=0 width=8)
>
> Sometimes I wish I had the power to tell the DBMS how I wanted a query
> done...

Yep, it is on our TODO list, and we have someone trying some fix for
6.4.  It has to do the conjunctive normal form(cnf).

--
Bruce Momjian                          |  830 Blythe Avenue
maillist@candle.pha.pa.us              |  Drexel Hill, Pennsylvania 19026
  +  If your life is a hard drive,     |  (610) 353-9879(w)
  +  Christ can be your backup.        |  (610) 853-3000(h)