Durable Tables
Data tables and other in-memory objects (thesauri, custom character maps, and Learn Models) can be automatically saved to a permanent storage device. This feature is called durable data. It's primary purpose is to permit efficient recovery of up-to-date data when the ibi Patterns - Search server is restarted after a planned or unplanned shutdown.
Some additional points to keep in mind about the durable data feature:
| • | Durable Data is automatic. |
| • | Durable Data supersedes the Checkpoint/Restore feature. When durable data is enabled, the checkpoint and restore commands do nothing. |
| • | Durable data persists all data: tables, thesauri, custom character maps, and Learn Models |
| • | Durable Data does not provide a history of data, only the current data. |
| • | Durable Data is safe. This means data changes are handled within a transaction system, and are immediately entrusted to the operating system. If the software crashes, the operating system is still able to update the persistent storage device. |
| • | Durable data is fast. It has only a modest impact on the speed of update operations. Very large updates (100 million records or more) must be done only when the system is inactive. |
| • | Durable Data is automatically reloaded when the server is restarted. |
| • | Deleting a table, thesaurus, or Learn Model deletes all associated durable data. |
| • | Durable data is NOT encrypted. The persistence directory, and any files in it, must be protected from unauthorized access. |
| • | On Windows, the Durable-Data feature imposes a limit of 100 tables. |
| • | On Linux, the Durable-Data feature imposes a limit on the number of tables based on the value reported by ulimit -n. Divide this value by 100 to obtain the limit on the number of tables. |
| • | If there are large tables (over 10M records), loading the data at server start-up can take noticeable time. Applications may encounter a timeout error if they connect to ibi Patterns - Search while it is loading large tables at start-up. |