>> i move info from support list what i provide there >> 1. I use Super Server 64 bit > > Certainly super does not write pages to disk as active as classic, but... > >> 2. DB Page Size is 16k and DefaultDBPageCache is 65536 (i disable system >> cache in fbconfig) > > that's just 1Gb cache. We GC pages during sweep in relation order, but > if indices of some relation(s) do not fit in the cache, pages will be > swapped in/out very active. > >> I do not know if this is important info but: >> l have ~4 indexes per table – but tables have not any constraints like pk, >> uk – >> and some indexes have depth 3, all others have depth 2 >> In database all records are small max 100 bytes - no blobs > >>> So the question is - does your DB have relations with size of indices > >>> 1Gb? > > Yes, some of indexes are bigger then 1GB > biggest have 4GB > > but why index must fit into cache? > and the bigger question why is written back to disc? > I understand that i can get more reads becouse of "small" cache but why > writes?
I haven't followed the entire thread, but if you are using Firebird 2.5.2 you could use the Trace API with log_sweep enabled to see I/O stats per table. Regards, Thomas ------------------------------------------------------------------------------ Master SQL Server Development, Administration, T-SQL, SSAS, SSIS, SSRS and more. Get SQL Server skills now (including 2012) with LearnDevNow - 200+ hours of step-by-step video tutorials by Microsoft MVPs and experts. SALE $99.99 this month only - learn more at: http://p.sf.net/sfu/learnmore_122512 Firebird-Devel mailing list, web interface at https://lists.sourceforge.net/lists/listinfo/firebird-devel