The original dataset includes DOJ_RECORD_ID and PERSON_NUMBER which uniquely identify the event and the people involved, respectively.
At the moment, for ripa-2018-db.herokuapp.com a UNIQUE_INDEX was created combining DOJ_RECORD_ID and PERSON_NUMBER. However, this is a 22 character string.
A numeric unique id would be much less costly in terms of memory. Both DOJ_RECORD_ID and PERSON_NUMBER would remain in the database in the base table (to be renamed). This unique id would only be used to connect tables uniquely identifying each row (even-person pair).
It could potentially be as simple as enumerating each row.
The original dataset includes
DOJ_RECORD_ID
andPERSON_NUMBER
which uniquely identify the event and the people involved, respectively.At the moment, for ripa-2018-db.herokuapp.com a
UNIQUE_INDEX
was created combiningDOJ_RECORD_ID
andPERSON_NUMBER
. However, this is a 22 character string.A numeric unique id would be much less costly in terms of memory. Both
DOJ_RECORD_ID
andPERSON_NUMBER
would remain in the database in the base table (to be renamed). This unique id would only be used to connect tables uniquely identifying each row (even-person pair).It could potentially be as simple as enumerating each row.