Question

我有一个带有自动递增列的postgresql数据库，我想迁移到Vertica。在Vertica中，我可以使用IDENTITY或AUTO INCREMENT约束（或数据类型？）创建一个表，但我无法将这些约束添加到包含现有数据的列。类似问题here的答案是使用序列的解决方法，这不是一个选项，因为允许重复的数字，并且可能存在多个节点之间的竞争条件，这由IDENTITY实现处理。使用IDENTITY列创建表也不是一种选择，因为不允许插入。

简而言之，我希望在现有数据上有一个IDENTITY列，同时也是主键。

Answer 1

您的问题是，您需要为Vertica中的副本中的原始IDENTITY列提供相同的值 - 同时能够对现在可能插入新表的任何新行具有AUTOINCREMENT行为。

这可以通过将Vertica目标代理键不是IDENTITY，而是作为INT NOT NULL，但是从手动创建的序列的nextval方法给它一个DEFAULT来实现。

你有这个表，这个数据（可以来自任何源数据库......），它运行IDENTITY列。

-- source (could be from anywhere - why not on my own Vertica sandbox)                                                                                      
CREATE TABLE src (
  src_id     IDENTITY NOT NULL
, first_name VARCHAR(32)
, last_name  VARCHAR(32)
, hire_dt    DATE
) UNSEGMENTED ALL NODES
;

INSERT /*+ DIRECT */ INTO src(first_name,last_name,hire_dt)
          SELECT 'Ford','Prefect',DATE '2017-02-05'
UNION ALL SELECT 'Svlad','Cjelli',DATE '2017-02-05'
UNION ALL SELECT 'Cynthia','Fitzmelton',DATE '2017-02-05'
UNION ALL SELECT 'Stavro','Mueller',DATE '2017-02-05'
UNION ALL SELECT 'Veet','Voojagig',DATE '2017-02-05'
UNION ALL SELECT 'Trin','Tragula',DATE '2017-02-05'
UNION ALL SELECT 'Zarniwoop','Zarniwoop',DATE '2017-02-05'
UNION ALL SELECT 'Rob','McKenna',DATE '2017-02-05'
UNION ALL SELECT 'The Lajestic Vantrashell','of Lob',DATE '2017-02-05'
UNION ALL SELECT 'Paul Neil Milne','Johnston',DATE '2017-02-05'
UNION ALL SELECT 'Lunkwill','Lunkwill',DATE '2017-02-05'
UNION ALL SELECT 'Arthur','Dent',DATE '2017-02-05'
UNION ALL SELECT 'Zaphod','Beeblebrox',DATE '2017-02-05'
UNION ALL SELECT 'Tricia','McMillan',DATE '2017-02-05'
UNION ALL SELECT 'Prostetnic Vogon','Jeltz',DATE '2017-02-05'
UNION ALL SELECT 'Lionel','Prosser',DATE '2017-02-05'
UNION ALL SELECT 'Karl','Mueller',DATE '2017-02-05'
UNION ALL SELECT 'Hotblack','Desiato',DATE '2017-02-05'
UNION ALL SELECT 'Gogrilla','Mincefriend',DATE '2017-02-05'
UNION ALL SELECT 'Slartibartfast','Slartibartfast',DATE '2017-02-05'
UNION ALL SELECT 'Roosta','Roosta',DATE '2017-02-05'
UNION ALL SELECT 'Eccentrica','Gallumbitis',DATE '2017-02-05'
UNION ALL SELECT 'Pizpot','Gargravarr',DATE '2017-02-05'
UNION ALL SELECT 'Vroomfondel','Vroomfondel',DATE '2017-02-05'
UNION ALL SELECT 'Majikthise','Majikthise',DATE '2017-02-05'
UNION ALL SELECT 'Gengis Temüjin','Khan',DATE '2017-02-05'
UNION ALL SELECT 'Know-Nothing-Bozo','the Non-Wonder Dog',DATE '2017-02-05'
UNION ALL SELECT 'Lazlaar','Lyricon',DATE '2017-02-05'
UNION ALL SELECT 'Lintilla','Lintilla',DATE '2017-02-05'
UNION ALL SELECT 'Fook','Fook',DATE '2017-02-05'
UNION ALL SELECT 'Gag','Halfrunt',DATE '2017-02-05'
UNION ALL SELECT 'Benji','Mouse',DATE '2017-02-05'
UNION ALL SELECT 'Frankie','Mouse',DATE '2017-02-05'
UNION ALL SELECT 'Grunthos','the Flatulent',DATE '2017-02-05'
UNION ALL SELECT 'Wowbagger','The Infinitely Prolonged',DATE '2017-02-05'
UNION ALL SELECT 'Wonko','The Sane',DATE '2017-02-05'
UNION ALL SELECT 'Reg','Nullify',DATE '2017-02-05'
UNION ALL SELECT 'Fenchurch','of Rickmansworth',DATE '2017-02-05'
UNION ALL SELECT 'Oolon','Colluphid',DATE '2017-02-05'
UNION ALL SELECT 'Humma','Kavula',DATE '2017-02-05'
UNION ALL SELECT 'Judiciary','Pag',DATE '2017-02-05'
UNION ALL SELECT 'Max','Quordlepleen',DATE '2017-02-05'
;

从该源表中，您可以确定当前发布的最高IDENTITY值：

SQL>select max(src_id) from src;
max
 42

在目标Vertica数据库中，您使用该值创建全新的手动序列...

CREATE SEQUENCE seq_copy MINVALUE 43;

使用代理ID创建目标表，具体取决于给定的输入值（您将在INSERT ... SELECT或COPY命令中提供）或新创建的序列{{1 }}：

nextval

然后，如果您指定所有列（或根本没有）

在INSERT中......

CREATE TABLE cpy (
  cpy_id     INT NOT NULL DEFAULT(seq_copy.nextval)
, first_name VARCHAR(32)
, last_name  VARCHAR(32)
, hire_dt    DATE
) UNSEGMENTED ALL NODES
;

..或COPY ......

INSERT INTO cpy (
  cpy_id
, first_name
, last_name
, hire_dt
)
SELECT * FROM src;

...您使用来自源的值。

如果您插入新值，则不要触摸代理键...

COPY cpy (cpy_id,first_name,last_name,hire_dt) 
  FROM LOCAL 'src.csv' DELIMITER ',' ENCLOSED BY '''' ;

...并且将为您填充代理键。

这是你之后的事吗？

马

将AUTO INCREMENT或IDENTITY添加到Vertica中的现有表

1 个答案: