RE: Spam:range partition issue
- From: "Ric Van Dyke" <ric.van.dyke@xxxxxxxxxx>
- To: "Bob Carlin" <smeghead.rimmer@xxxxxxxxx>
- Date: Thu, 15 Nov 2007 10:30:52 -0600
Roger that. I was a bit too quick in my response.
Here is a test that shows what I was trying (but messed up) to say
yesterday:
-------------------------------------------------------------------
SQL> r
1 CREATE TABLE test_part
2 (
3 Col1 NUMBER(10) NOT NULL,
4 Col2 NUMBER(10) NOT NULL
5 )
6 PARTITION BY RANGE (COL1, COL2)
7 (
8 PARTITION USAGE_1 VALUES LESS THAN (897515520, 1500000),
9 PARTITION USAGE_2 VALUES LESS THAN (897515520, 3500000)
10* )
SQL> insert into test_part values(897515520,1200000);
SQL> insert into test_part values(897515520,1600000);
SQL>
SQL> select count(*) from test_part partition(usage_1);
COUNT(*)
---------------
1
SQL> select count(*) from test_part partition(usage_2);
COUNT(*)
---------------
1
SQL>
*********** Data is in the partitions as expected.
SQL>
SQL> insert into test_part values(897515519,1600000);
*********** This insert "should" be in the second partition.
SQL> insert into test_part values(897515519,1200000);
*********** This insert should be in the first partition.
SQL> select count(*) from test_part partition(usage_1);
COUNT(*)
---------------
3
*********** Both are really in the first.
SQL> select count(*) from test_part partition(usage_2);
COUNT(*)
---------------
1
SQL>
SQL> select * from test_part partition(usage_1);
COL1 COL2
--------------- ---------------
897515520 1200000
897515519 1600000
897515519 1200000
-----------------------
Ric Van Dyke
Hotsos Enterprises
Cell 248-705-0624
-----------------------
-----Original Message-----
From: Bob Carlin [mailto:smeghead.rimmer@xxxxxxxxx]
Sent: Thursday, November 15, 2007 11:21 AM
To: Ric Van Dyke
Cc: Harvinder.Singh@xxxxxxxxxxxxx; oracle-l@xxxxxxxxxxxx;
oracle-l@xxxxxxxxxxxxx
Subject: Re: Spam:range partition issue
Ric,
Actually is works by comparing until is does *not* have a match. I
tested Harvinder's example, and it behaves as expected; I think
Harvinder must have had an error in his own test.
Quoting from the manual:
"When a table or index is partitioned by range on multiple columns, each
partition bound and partitioning key is a list (or vector) of values.
The partition bounds and keys are ordered according to ANSI SQL2 vector
comparison rules. This is also the way Oracle orders multicolumn index
keys.
To compare a partitioning key with a partition bound, you compare the
values of their corresponding columns until you find an unequal pair and
then that pair determines which vector is greater. The values of any
remaining columns have no effect on the comparison."
Ric Van Dyke wrote:
>
> Oracle compares from left to right and stops when it has a match:
>
> First row: 897515520,1200000 The first value is less then or equal to
> the first key of the partition, so it goes in that partition.
>
> Second row: 897515520,1600000 The first value is less then or equal to
> the first key of the partition, so it goes in that partition.
>
> In both cases it never even looks at the second value. I forget the
> name of this mechanism but it's the way it works.
>
> -----------------------
>
> Ric Van Dyke
>
> Hotsos Enterprises
>
> -----------------------
>
>
------------------------------------------------------------------------
>
> *From:* oracle-l-bounce@xxxxxxxxxxxxx
> [mailto:oracle-l-bounce@xxxxxxxxxxxxx] *On Behalf Of *Harvinder Singh
> *Sent:* Wednesday, November 14, 2007 3:37 PM
> *To:* oracle-l@xxxxxxxxxxxx; oracle-l@xxxxxxxxxxxxx
> *Subject:* Spam:range partition issue
>
> Hi,
>
> I am creating range partitioning on 2 columns as following:
>
> CREATE TABLE test_part
>
> (
>
> Col1 NUMBER(10) NOT NULL,
>
> Col2 NUMBER(10) NOT NULL
>
> )
>
> TABLESPACE tab_1
>
> PARTITION BY RANGE (ID_USAGE_INTERVAL, ID_ACC)
>
> (
>
> PARTITION USAGE_1 VALUES LESS THAN (897515520, 1500000)
>
> TABLESPACE tab_2,
>
> PARTITION USAGE_2 VALUES LESS THAN (897515520, 3500000)
>
> TABLESPACE tab_3
>
> )
>
> insert into test_part values(897515520,1200000)
>
> insert into test_part values(897515520,1600000)
>
> commit
>
> select count(*) from test_part partition(usage_1)
>
> select count(*) from test_part partition(usage_2)
>
> I expect both the partitions usage1 and usage2 to have 1 row each but
> both the rows are going to usage1, What can be the possible issue?
>
> Thanks
>
> --Harvinder
>
--
//www.freelists.org/webpage/oracle-l
Other related posts:
