On Sun, 3 Nov 2003, Tim Churches wrote: ... > > 2) restore (converts /home/knoppix/publicdb.zip back into publicdb in > > PostgreSQL) > > Are you just zipping the PG data directories?
Tim, No, we are using pg_dump. > Or are you dumping the database to a separate file and zipping that? Yes, exactly. This is what we are doing for now. In the future, we will have incremental backup. We already have a single table that tracks all patient "records" being added into the system. We can use that table to track backups. ... > > An interesting implication is that it is now possible to build OIO > > production servers that run entirely from CD-ROM and RAM - with periodic > > auto-backup to networked storage or solid state memory, like USB connected > > flash drives. > > You would need to use dump to do this, which allows online backup while > the database is still running. PG dump is very fast, so it would be > feasible to backup a RAM database every minute or so, provided teh > database was not too large. Ideally, we should move to incremental backup for database that contain more than 10000 patients (about 50MB). It is not difficult to add this capability but one step at a time. Best regards, Andrew --- Andrew P. Ho, M.D. OIO: Open Infrastructure for Outcomes www.TxOutcome.Org
