Why would we create a sequence even if there is a primary key?
The primary key is a column in a table.
The primary key needs a unique value, which needs to come from somewhere.
The sequence is a feature by some database products which just creates unique values. It just increments a value and returns it. The special thing about it is: there is no transaction isolation, so several transactions can not get the same value, the incrementation is also not rolled back. Without a database sequence, it is very hard to generate unique incrementing numbers.
Other database products support columns that are automatically initialized with an incrementing number.
There are other means to create unique values for the primary keys, for instance Guids.
Sequence will allow you to populate primary key with a unique, serialized number.
It's different from a serial
or auto_incremement
primary key in the sense that:
It is an actual database object (you need to create it):
sql> create sequence NAME_OF_YOUR_SEQUENCE;
You could assign independent permissions to it, to different database users:
sql> grant select on NAME_OF_YOUR_SEQUENCE to NAME_OF_YOUR_USER;
You can use to have a unique number that is different among several tables (not just one). Say you have four tables with numeric primary keys, and you want unique numbers among those four tables. You could use a sequence for that, without having to worry about implementing locking mechanisms to do it 'by hand'.
You can change its number to any value you want with
alter sequence
You can cycle through its numbers
sql> create sequence NAME_OF_YOUR_SEQUENCE maxvalue 1500 cycle;
The primary key is (in technical terms) merely an index that enforces uniqueness (as well as speeding query performance). There's some semantic information there along that being the "key" for the entity the row describes, but that's it.
A sequence is a different entity entirely; it exists separate from tables (like a stored procedure would) and can be called to yield sequential numbers.
The two are often used together, to generate automatic primary keys for entities that have no sensible "native" keys. But they are two separate concepts; you can have tables where the primary key is explicitly populated during an insert, and you can have sequences that are used to populate non-PK columns (or even used imperatively during a stored procedure, distinct from inserting records).
© 2022 - 2024 — McMap. All rights reserved.