CTIDs are a field in every PostgreSQL table and are known to identify specific physical rows according to their block and offset positions within a particular table. They are used by index entries to point to physical rows. It is unique for each record in the table and quickly denotes the location of a tuple.
A logical row’s CTID changes when updated, so the CTID cannot be used as a long-term row identifier. However, it is sometimes helpful to identify a row within a transaction when no competing update is expected.
