miconda left a comment (kamailio/kamailio#4306)
A few questions to understand the design for the new struct storage model: did I understand correctly that a DB record (ie., row values) are now keys in a dedicated hash table (since the value is `DUMMY`)? If yes, they are prefixed with the column name, or how is the mapping column-name to hash-key (the value) is done? Afaik, uasualy the order of the items/keys in a hash table is not guaranteed, or does Redis guarantees that for the same list of keys, it is the same order of the items?
Where per-item expiration helps (e.g., for usrloc), isn't the entire hash table that has to be deleted on expire? Or is not like each DB record is a dedicated htable?
Sorry for the laziness of not going through the code to understand from there some of the questions ...