High Performance considerations

1. Table data to maintain a certain number of commonly used tables consider data dumps, do not let the volume of data is too large (commercial one-time login, registration failure, is because the Deviceonline table data reached millions, the association time is too long resulting in timeouts; Deviceonline is actually designed to be deleted every hour, Timed task failure; however, as a data analysis reserve, the data should not be deleted, but should be dumped, when the dump can be processed, rather than intact, such as the Deviceonline dump, you can eliminate the history of online records, only dump the difference of data, etc.);
2. Only inserted, not updated; ibid., Deviceonline mechanism;
3. Filtering mechanism, early equipment escalation, resulting in excessive server pressure, added a filtering mechanism, if the escalation status and the same, no longer go to business logic processing (unpacking, analysis, plug-in database, etc.);

High Performance considerations