Currently the entire data is inserted into the database in one big transaction. This only works because only a tiny fraction of the data is used. Someone really needs to find a way to import data in chunks. One major obstacle is to connect these chunks via edges because then the actual node ids would have to be read back.
Currently the entire data is inserted into the database in one big transaction. This only works because only a tiny fraction of the data is used. Someone really needs to find a way to import data in chunks. One major obstacle is to connect these chunks via edges because then the actual node ids would have to be read back.