Logo Questions Linux Laravel Mysql Ubuntu Git Menu
 

how to partition a table by datetime column?

I want to partition a mysql table by datetime column. One day a partition.The create table scripts is like this:

CREATE TABLE raw_log_2011_4 (
  id bigint(20) NOT NULL AUTO_INCREMENT,
  logid char(16) NOT NULL,
  tid char(16) NOT NULL,
  reporterip char(46) DEFAULT NULL,
  ftime datetime DEFAULT NULL,
  KEY id (id)
) ENGINE=InnoDB AUTO_INCREMENT=286802795 DEFAULT CHARSET=utf8
PARTITION BY hash (day(ftime)) partitions 31;

But when I select data of some day.It could not locate the partition.The select statement is like this:

explain partitions select * from raw_log_2011_4 where day(ftime) = 30;

when i use another statement,it could locate the partition,but I coluld not select data of some day.

explain partitions select * from raw_log_2011_4 where ftime = '2011-03-30';

Is there anyone tell me How I could select data of some day and make use of partition.Thanks!

like image 666
tinychen Avatar asked May 23 '11 06:05

tinychen


People also ask

Can you partition by date in SQL?

You can partition on DateCreated column after you make it a clustered index.

How do I partition a table in SQL?

To create a partitioned table, you follow these steps: Create file groups that hold the partitions of the table. Create a partition function that maps the rows of the table into partitions based on the values of a specified column. Create a partition scheme that maps the partition table to the new filegroups.


2 Answers

Partitions by HASH is a very bad idea with datetime columns, because it cannot use partition pruning. From the MySQL docs:

Pruning can be used only on integer columns of tables partitioned by HASH or KEY. For example, this query on table t4 cannot use pruning because dob is a DATE column:

SELECT * FROM t4 WHERE dob >= '2001-04-14' AND dob <= '2005-10-15';

However, if the table stores year values in an INT column, then a query having WHERE year_col >= 2001 AND year_col <= 2005 can be pruned.

So you can store the value of TO_DAYS(DATE()) in an extra INTEGER column to use pruning.

Another option is to use RANGE partitioning:

CREATE TABLE raw_log_2011_4 (
  id bigint(20) NOT NULL AUTO_INCREMENT,
  logid char(16) NOT NULL,
  tid char(16) NOT NULL,
  reporterip char(46) DEFAULT NULL,
  ftime datetime DEFAULT NULL,
  KEY id (id)
) ENGINE=InnoDB AUTO_INCREMENT=286802795 DEFAULT CHARSET=utf8
  PARTITION BY RANGE( TO_DAYS(ftime) ) (
    PARTITION p20110401 VALUES LESS THAN (TO_DAYS('2011-04-02')),
    PARTITION p20110402 VALUES LESS THAN (TO_DAYS('2011-04-03')),
    PARTITION p20110403 VALUES LESS THAN (TO_DAYS('2011-04-04')),
    PARTITION p20110404 VALUES LESS THAN (TO_DAYS('2011-04-05')),
    ...
    PARTITION p20110426 VALUES LESS THAN (TO_DAYS('2011-04-27')),
    PARTITION p20110427 VALUES LESS THAN (TO_DAYS('2011-04-28')),
    PARTITION p20110428 VALUES LESS THAN (TO_DAYS('2011-04-29')),
    PARTITION p20110429 VALUES LESS THAN (TO_DAYS('2011-04-30')),
    PARTITION future VALUES LESS THAN MAXVALUE
  );

Now the following query will only use partition p20110403:

SELECT * FROM raw_log_2011_4 WHERE ftime = '2011-04-03';
like image 77
Steyx Avatar answered Sep 27 '22 20:09

Steyx


Hi You are doing the wrong partition in definition of the table the table definition would like this:

CREATE TABLE raw_log_2011_4 (
  id bigint(20) NOT NULL AUTO_INCREMENT,
  logid char(16) NOT NULL,
  tid char(16) NOT NULL,
  reporterip char(46) DEFAULT NULL,
  ftime datetime DEFAULT NULL,
  KEY id (id)
) ENGINE=InnoDB AUTO_INCREMENT=286802795 DEFAULT CHARSET=utf8
PARTITION BY hash (TO_DAYS(ftime)) partitions 31;

And your select command would be:

explain partitions 
    select * from raw_log_2011_4 where TO_DAYS(ftime) = '2011-03-30';

The above command would select all the date required, as if you use the TO_DAYS command as

mysql> SELECT TO_DAYS(950501);
        -> 728779
mysql> SELECT TO_DAYS('2007-10-07');
        -> 733321

Why to use the TO_DAYS AS The MySQL optimizer will recognize two date-based functions for partition pruning purposes: 1.TO_DAYS() 2.YEAR()

and this would solve your problem..

like image 41
Vineet1982 Avatar answered Sep 27 '22 22:09

Vineet1982