Re: Primary key type selection

ahhhh, free time to experiment, i'd never tried it, but didnt think it would make enough of a diff to warrant one over the other.

thanks, joe

Jared Still wrote:

> Here's a quick and dirty test on this. The actual number
> of rows in the test table was 960,000.
>
> First is the create script, then the test script, then the
> results. The differences in time are not worth worrying about.
>
> Jared
>
> ---------------------------------------
>
> drop table pk_test;
> drop sequence pk_test_seq;
>
> create table pk_test (
> uid1 number(9,0) not null,
> uid2 varchar2(9) not null,
> data date default sysdate not null
> )
> storage ( initial 1m next 1m pctincrease 0 )
> /
>
> create sequence pk_test_seq start with 1000000;
>
> create or replace trigger pk_test_bit
> before insert on pk_test
> for each row
> declare
> newPk integer;
> begin
> select pk_test_seq.nextval into newPK
> from dual;
>
> :new.uid1 := newPk;
> :new.uid2 := to_char(newPk);
>
> end;
> /
>
> declare
> v_commit_point integer := 10000;
> v_commit integer := 0;
> begin
> for x in 1 .. 1000000
> loop
> insert into pk_test(data) values( sysdate );
> v_commit := v_commit + 1;
> if v_commit >= v_commit_point then
> commit;
> v_commit := 0;
> end if;
> end loop;
> commit;
> end;
> /
>
> alter table pk_test add constraint uid1 unique ( uid1 )
> using index
> tablespace indx
> storage ( initial 1m next 1m pctincrease 0 )
> /
>
> alter table pk_test add constraint uid2 unique ( uid2 )
> using index
> tablespace indx
> storage ( initial 1m next 1m pctincrease 0 )
> /
>
> ---------------------------------------
>
> set autotrace on
> set timing on
>
> select /*+ index(pk_test,uid1) */ count(uid1) from pk_test;
> select /*+ index(pk_test,uid2) */ count(uid2) from pk_test;
>
> set timing off
> set autotrace off
>
> ---------------------------------------
>
> Segments for jkstill
> COUNT(UID1)
> -----------
> 960000
>
> 1 row selected.
>
> Elapsed: 00:00:02.40
>
> Execution Plan
> ----------------------------------------------------------
> 0
> SELECT STATEMENT Optimizer=CHOOSE (Cost=26 Card=296401 Bytes=3853213)
>
> 1 0
> SORT (AGGREGATE)
>
> 2 1
> INDEX (FULL SCAN) OF 'UID1' (UNIQUE) (Cost=26 Card=296401 Bytes=3853213)
>
> Statistics
> ----------------------------------------------------------
> 8 recursive calls
> 0 db block gets
> 4388 consistent gets
> 4382 physical reads
> 0 redo size
> 248 bytes sent via SQL*Net to client
> 307 bytes received via SQL*Net from client
> 4 SQL*Net roundtrips to/from client
> 0 sorts (memory)
> 0 sorts (disk)
> 1 rows processed
>
> Segments for jkstill
> COUNT(UID2)
> -----------
> 960000
>
> 1 row selected.
>
> Elapsed: 00:00:02.51
>
> Execution Plan
> ----------------------------------------------------------
> 0
> SELECT STATEMENT Optimizer=CHOOSE (Cost=26 Card=296401 Bytes=1778406)
>
> 1 0
> SORT (AGGREGATE)
>
> 2 1
> INDEX (FULL SCAN) OF 'UID2' (UNIQUE) (Cost=26 Card=296401 Bytes=1778406)
>
> Statistics
> ----------------------------------------------------------
> 8 recursive calls
> 0 db block gets
> 4936 consistent gets
> 4926 physical reads
> 0 redo size
> 248 bytes sent via SQL*Net to client
> 307 bytes received via SQL*Net from client
> 4 SQL*Net roundtrips to/from client
> 0 sorts (memory)
> 0 sorts (disk)
> 1 rows processed
>
> ---------------------------------------
>
> On Tue, 16 May 2000, Joseph S. Testa wrote:
>
> > Winnie, u've tested that theory that a number scan as a PK will be faster than
> > a character scan(varchar2)?, i'd like to see the results :)
> >
> > joe
> >
> >
> > Winnie Liu wrote:
> >
> > > Colin,
> > >
> > > That's right. The number field occupied varies as varchar2. But still, with
> > > the same width, it is a lot faster to scan and match a primary key of
> > > numbers than a primary key with characters.
> > >
> > > Winnie :)
> > >
> > > -----Original Message-----
> > > Colin.Shaw_at_phs.com
> > > Sent: Monday, May 15, 2000 9:28 PM
> > > To: Multiple recipients of list ORACLE-L
> > >
> > > >A number field takes less no. of bytes than a varchar2 field.
> > >
> > > Not necessarily. A number field occupies a variable number of bytes,
> > > depending on its magnitude and number of digits.
> > >
> > > Colin.
> > >
> > > -----Original Message-----
> > > Sent: Sunday, May 14, 2000 9:14 PM
> > > To: Multiple recipients of list ORACLE-L
> > >
> > > Hi,
> > >
> > > The first advantage is the space saving. A number field takes less no.
> > > of bytes than a varchar2 field. This decrease in storage may
> > > positively impact performance...
> > >
> > > Regards
> > > Rajagopal Venkataramany
> > >
> > > ----Original Message Follows----
> > > Reply-To: ORACLE-L_at_fatcity.com
> > > To: Multiple recipients of list ORACLE-L <ORACLE-L_at_fatcity.com>
> > > Date: Sun, 14 May 2000 10:54:12 -0800
> > >
> > > Is there any advantages and disadvantages (performance? , ...) between
> > > NUMBER and VARCHAR2 for primary key column?
> > >
> > > Alex Hillman
> > >
> > > ________________________________________________________________________
> > > Get Your Private, Free E-mail from MSN Hotmail at http://www.hotmail.com
> > >
> > > --
> > > Author: Rajagopal Venkataramany
> > > INET: rajagopalvr_at_hotmail.com
> > >
> > > Fat City Network Services -- (858) 538-5051 FAX: (858) 538-5051
> > > San Diego, California -- Public Internet access / Mailing Lists
> > > --------------------------------------------------------------------
> > > To REMOVE yourself from this mailing list, send an E-Mail message
> > > to: ListGuru_at_fatcity.com (note EXACT spelling of 'ListGuru') and in
> > > the message BODY, include a line containing: UNSUB ORACLE-L
> > > (or the name of mailing list you want to be removed from). You may
> > > also send the HELP command for other information (like subscribing).
> > >
> > > --
> > > Author:
> > > INET: Colin.Shaw_at_phs.com
> > >
> > > Fat City Network Services -- (858) 538-5051 FAX: (858) 538-5051
> > > San Diego, California -- Public Internet access / Mailing Lists
> > > --------------------------------------------------------------------
> > > To REMOVE yourself from this mailing list, send an E-Mail message
> > > to: ListGuru_at_fatcity.com (note EXACT spelling of 'ListGuru') and in
> > > the message BODY, include a line containing: UNSUB ORACLE-L
> > > (or the name of mailing list you want to be removed from). You may
> > > also send the HELP command for other information (like subscribing).
> > >
> > > --
> > > Author: Winnie Liu
> > > INET: hkitty_at_earthlink.net
> > >
> > > Fat City Network Services -- (858) 538-5051 FAX: (858) 538-5051
> > > San Diego, California -- Public Internet access / Mailing Lists
> > > --------------------------------------------------------------------
> > > To REMOVE yourself from this mailing list, send an E-Mail message
> > > to: ListGuru_at_fatcity.com (note EXACT spelling of 'ListGuru') and in
> > > the message BODY, include a line containing: UNSUB ORACLE-L
> > > (or the name of mailing list you want to be removed from). You may
> > > also send the HELP command for other information (like subscribing).
> >
> > --
> > Author: Joseph S. Testa
> > INET: teci_at_oracle-dba.com
> >
> > Fat City Network Services -- (858) 538-5051 FAX: (858) 538-5051
> > San Diego, California -- Public Internet access / Mailing Lists
> > --------------------------------------------------------------------
> > To REMOVE yourself from this mailing list, send an E-Mail message
> > to: ListGuru_at_fatcity.com (note EXACT spelling of 'ListGuru') and in
> > the message BODY, include a line containing: UNSUB ORACLE-L
> > (or the name of mailing list you want to be removed from). You may
> > also send the HELP command for other information (like subscribing).
Received on Wed May 17 2000 - 12:13:54 CDT