Here, the producer informs the consumer of data validity, where timing is inferred from data validity.
So a data_valid flag is used as a clock in this case where sample times are linked to data_valid going high.
Performance
These are usually better for performance but could introduce things like replication lag and eventual consistency.