Re:Ask HN: GUID, int, or both for primary key?
by daleholborow on Wednesday May 09, 2018 @01:51AM
(#17027659)
There are additional arguments against using ints as ids in the database, if you ever plan to shard etc, as in "you'd still have to do a lot of work to migration data from an integer-key-based-system to something distributed". Various articles describe this.
Keep in mind that many RDBMS systems now support some concept of a "sequential guid" - but keep in mind that they all store these in different ways.
Plug: here's some code I pulled together, with links to the reference material, that will generate sequential guids for various databases:
Speaking from personal experience, in theory I would always choose a (sequential)GUID as PK.. BUT... they ARE a pain in the bollocks to work with when you are developing and writing sql by hand to query data. E.g.
a) select * from Thingies where Id = 5
vs
b) select * from Thingies where Id = '12331-432Aafds-32424-df'
https://github.com/daleholborow/iayos.sequentialguid
Speaking from personal experience, in theory I would always choose a (sequential)GUID as PK.. BUT... they ARE a pain in the bollocks to work with when you are developing and writing sql by hand to query data. E.g.
a) select * from Thingies where Id = 5
vs
b) select * from Thingies where Id = '12331-432Aafds-32424-df'