Full unicode support conversion impractically slow

Testing must be done on a MySQL (or MariaDB) database. The collation for your database needs to be three byte UTF-8 (e.g. utf8_unicode_ci). Create two large courses using the test course creator. (Site administration > Development > make test course) Make one of the tables in the database use the 'Compact' row format. You can update a table to use 'Compact' with the following sql -  ALTER TABLE `mdl_auth_oauth2_linked_login` ROW_FORMAT=COMPACT Use the following sql to check that row format SELECT row_format FROM information_schema.tables WHERE table_schema = {name of your database here} AND table_name = 'mdl_auth_oauth2_linked_login' To get an idea of how long the script took before make the following changes: line 122 $sql = "ALTER DATABASE $CFG->dbname CONVERT TO CHARACTER SET $charset COLLATE $collation"; back to $sql = "ALTER DATABASE $CFG->dbname DEFAULT CHARACTER SET $charset DEFAULT COLLATE = $collation"; Add some additional code to check the time taken to run the script. Run the CLI script php admin/cli/mysql_collation collation=utf8mb4_unicode_ci Make note of how long it took to run the script. Revert the database back to utf8_unicode_ci. Change the script to use the CONVERT TO query. Run the script again and note the time taken to finish. The new change to the query should show a significant difference in time taken to execute. Run the sql query to check the row format of the previously 'Compact' table to ensure that it has been changed to 'Compressed'