From: Rory Campbell-L. <ro...@ca...> - 2011-07-21 19:57:05
|
On 21/07/11, Eric Bollengier (eri...@ba...) wrote: > Hello, > > >> When you add new indexes on the File table it leads to support > >> problems where people are complaining about backup speed... > > > > At least on this box, the CPU isn't an issue, so updating some more > > indices won't hurt too much. With some old box with old spinning rust > > for the databse, it might matter more, though. ... > If you want, a better solution would be to modify the query to run it in > two or three times and retrieve jobids first, then pathid/filenameid, > but it requires more complex code. (perhaps you can do tests on your > database and report us if you have any improvements) I think these are all rather interesting suggestions, because the core of Bacula is the database while Bacula itself has a rather offhand relationship with the database. Two examples: configuration could be easily stored in the database, and the tables aren't fully normalised. Perhaps there is scope for a side-project for the development of a set of Postgresql stored procedures and other items (cross-tables, indexes) suitable for larger backup catalogues? The benefits could be considerable -- faster performance and more command-line productivity for sysadmins; easier web client interfacing for those who need that. Thoughts? Rory -- Rory Campbell-Lange ro...@ca... Campbell-Lange Workshop www.campbell-lange.net 0207 6311 555 3 Tottenham Street London W1T 2AF Registered in England No. 04551928 |