Conversion from non partitioned table to partitioned table and precautions

Sometimes data archiving is not considered when designing at the beginning . Data archiving is not a simple export delete. Because our databases are generally delete Don't free up space . But we usually encounter insufficient space in the database , We have to face this problem . Of course, you can also do space recycling for defragmentation . I don't like it very much , Especially when it's a big watch .MySQL Once you surpass 100G Of , I dare not do it in online environment .Oracle I have done online recycling of 20T Table of , debris 800G, Recycled once and made it 55 Hours （ No downtime ）

Personally, I think the best way is to use zoning , It's easy to say whether to migrate partitions or archive partitions . Generally, the partition is truncated , Soon . But what if a table has not been partitioned before ？

Let's give an example

create table b (id int ,a varchar2(10),time date); Common heap table 

Copy

simulation 5 Data

insert into b values (1,'a',sysdate-120);

insert into b values (2,'b',sysdate-90);

insert into b values (3,'c',sysdate-60);

insert into b values (4,'d',sysdate-30);

insert into b values (5,'e',sysdate);

Copy

Generally speaking, we have to use time . So time indexing .

create index b1 on b (time);

Copy

Generally, tables have primary keys , stay Oracle The business document number should be used as the primary key , stay MySQL of use ID Self increment has no business meaning . stay PG According to the in Oracle Just come .

Alter table b add constraint pkb primary key(id);

Copy

by B Add primary key to the table , Note here that it must be named by itself rather than ordered by the database itself . Because there is no problem with normal masters and slaves , But there are OGG In this case, it is a pit . Later I'll talk about why .

ALTER TABLE b MODIFY

PARTITION BY RANGE (time)

(

PARTITION b1 VALUES LESS THAN (TO_DATE ('2022-03-01', 'YYYY-MM-DD')),

PARTITION b2 VALUES LESS THAN (TO_DATE ('2022-04-01', 'YYYY-MM-DD')),

PARTITION b3 VALUES LESS THAN (TO_DATE ('2022-05-01', 'YYYY-MM-DD')),

PARTITION b4 VALUES LESS THAN (TO_DATE ('2022-06-01', 'YYYY-MM-DD')),

PARTITION b5 VALUES LESS THAN (TO_DATE ('2022-07-01', 'YYYY-MM-DD'))

) online

UPDATE INDEXES

(

pkb GLOBAL,

b1 LOCAL

);

Copy

This is the command to convert non partitioned tables into separate tables online . Notice that in the red part, I specify the primary key name , If it is automatic, it is the name of automatic . The pit mentioned above is right here , Because in OGG scenario , The automatic name of the target side, that is, the downstream database, is different from that of the upstream side of the source side . Executing this command will cause OGG The interrupt .

Is this a database problem or OGG The problem of ？ I don't think it's , It's about getting familiar with the product . After execution, you can see that the table has become partitioned .

At this time, the expired data will be truncated as long as drop partition That's all right. , Of course, pay attention to one UPDATE GLOBAL INDEXES; Otherwise, the data cannot be written .

The source code attachment has been packaged and uploaded to Baidu cloud , You can download it yourself ～

link : https://pan.baidu.com/s/14G-bpVthImHD4eosZUNSFA?pwd=yu27 Extraction code : yu27 Baidu cloud link is unstable , It may fail at any time , Let's keep it tight .

If Baidu cloud link fails , Please leave me a message , When I see it, I will update it in time ～

Open source address Code cloud address ： http://github.crmeb.net/u/defu

Github Address ： http://github.crmeb.net/u/defu