Sqlite3 performance differences between the file database and the in-memory database

It's not a convincing test, it's my experience.

Through the Pandas.read_csv function, the total of 1.44GB of 164 CSV files in turn (into Dataframe object, and add 2 series, and then the dataframe through the Pandas.to_ The SQL function is written as a 1.23GB SQLite database file, and since the fourth file is imported, it is becoming slower and less expensive, with a total of two hours.

and Sqlite3.connect (': Memory: ') is only 2 minutes to write to the database, of course, not yet dumped to disk, I believe the time to dump should not be too much.

Sqlite3 performance differences between the file database and the in-memory database