From the SERIALIZABLE explanation: “The database runs the queries one by one … It is essential to have some retry mechanism since queries can fail.”
I know they’re trying to simplify, but this is confusing. If the first part is true, the second part can’t be. In reality the database does execute the queries concurrently, but will try to make it seem like they were done one by one. If it can’t manage that, a query will fail and have to be retried by the application.
I believe there was a caveat around this exact point later in the post. It was really tough striking a balance for people learning this for the first time and more knowledgeable audience without confusing them further.
I do appreciate the feedback and will look to add some more color here! Thank you!
I know they’re trying to simplify, but this is confusing. If the first part is true, the second part can’t be. In reality the database does execute the queries concurrently, but will try to make it seem like they were done one by one. If it can’t manage that, a query will fail and have to be retried by the application.