Re: Incremental backups, and backup history
От | Shridhar Daithankar |
---|---|
Тема | Re: Incremental backups, and backup history |
Дата | |
Msg-id | 3EF1C6BE.10966.AC878D8@localhost обсуждение исходный текст |
Ответ на | Incremental backups, and backup history (Antonios Christofides <A.Christofides@itia.ntua.gr>) |
Список | pgsql-general |
On 19 Jun 2003 at 11:42, Antonios Christofides wrote: > Hi, > > I have two backup questions, not much related to each other; here they > are. > > First: With PostgreSQL, I can't do incremental backups. pg_dump will > dump the entire database. Thus, if I want to keep daily backups on tape, > I'm in trouble because I'll have to do a full backup every day, which > may need several hours and several tapes. > > One workaround I'm thinking is to not store BLOBs in the database, but > store them in the filesystem and store the filenames in the database > instead. This needs some additional work (for example, a garbage > collector to periodically delete unreferenced files), but will move a > large amount of space from the database into the filesystem, which is > capable of incremental backups. Only BLOBs, that is; for some tables > that will have several tens of millions of small rows, I can't think of > any workaround. A cheaper solution in terms of efforts is to add a latest update timestamp to each blob and dump with a custom SQL. Incremental backups would be real nice. If you have real high turnover, I would suggest you to look asynchornous replication solution which work based on WAL sengmenting. That way you would backup only things that are changed. HTH Bye Shridhar -- Blore's Razor: Given a choice between two theories, take the one which is funnier.
В списке pgsql-general по дате отправления: