Rationale
I created an ETL in Python that downloads some data from Salesforce and loads them in SQL Server.
Those records are related to Salesforce Objects like Case, EmailMessage, RecordType, User, etc...
Those records also have a specific ID (varchar(18)
) that comes from Salesforce and it's used in the DB to JOIN (for example EmailMessage to Case) and it's also currently used as a PK.
Question
Will it make more sense, from the perspective of performance and best practices, to change the PK to an IDENTITY
column and just indexing the "salesforce id" columns and treat them just for JOIN
ing?