Hi. I expect I have some basic misunderstanding because IMO now this thread seems to have come full circle.
Earlier, Osumi-san was rejecting the idea of using ALTER TABLE tbl SET UNLOGGED on basis that it is too time consuming for large data to switch the table modes [1]. Now the latest idea is to introduce a wal_level=none. But now apparently full daily backups are OK, and daily restarting the server before the copies is also OK [2]. ~ Doesn't wal_level=none essentially just behave as if every table was UNLOGGED; not just the ones we are loading? Doesn't wal_level=none come with all the same limitations/requirements (full daily backups/restarts etc) that the UNLOGGED TABLE would also have? So I don't recognise the difference? If wal_level=none is judged OK as a fast loading solution, then why wasn't an initially UNLOGGED table also judged OK by the same criteria? And if there is no real difference, then why is it necessary to introduce wal_level=none (instead of using the existing UNLOGGED feature) in the first place? Or, if all this problem is simply due to a quirk that the BI tool referred to does not support the CREATE UNLOGGED TABLE syntax [3], then surely there is some other workaround could be written to handle that. What part am I missing? -- [1] - https://www.postgresql.org/message-id/OSBPR01MB48884832932F93DAA953CEB9ED650%40OSBPR01MB4888.jpnprd01.prod.outlook.com [2] - https://www.postgresql.org/message-id/TYAPR01MB299005FC543C43348A4993FDFE550%40TYAPR01MB2990.jpnprd01.prod.outlook.com [3] - https://www.postgresql.org/message-id/OSBPR01MB4888CBD08DDF73721C18D2C0ED790%40OSBPR01MB4888.jpnprd01.prod.outlook.com Kind Regards, Peter Smith. Fujitsu Australia