distributed

In a previous post, I explained that a serializable isolation level is not the only solution to avoid write skew anomalies. Normalizing the database schema, I converted the read-write conflict into a write-write conflict. This strategy allows us to achieve serialization within the read-committed isolation level using a SELECT FOR UPDATE.I used Martin Kleppmann's example of doctors' on-call shifts to demonstrate write skew anomalies. Document data modeling is often confused with denormalization, but this is inaccurate. I will store the doctors' on-call shifts in a JSONB document in this post. Using a single table as a document database allows me…