Search results for query: *

Dagon, thx for your support. there is a lot to try and I'm not sure which way I will go. I think the best way is to separate session data and to work with materialized views on historical data. but we will see. also if someone is interested in tuning measurements - this is a good link (in...

there is another index I created : CREATE INDEX "EDM"."IDX_OBJSAT_COMB_OBJID" ON "EDM"."SESSION_SELLER_TBL" ("SESSIONS_ID", "SELLER_ID") I had already an concatenated index on these 2 columns of sessions_tbl CREATE UNIQUE INDEX "EDM"."SESSIONS_STARTT_COMP_IDX" ON "EDM"."SESSIONS_TBL"...

I'm already using an index on any column but unfortunaly indexes on sessions_id or starttime are using only one key per block. Means there is no big performance advantage from indexes. I read a lot of arcticles today and it seems that the biggest advantage lays in partitioning of tables...

if I restrict the query by duration > 60 its returning 15.000 lines, without its returning 2.700.000 lines. Because the results are shown by jsp/html sites I need only the latest 3000 entries. similar to : select * from ( select * from sessions_vw order by starttime ) where rownum < 3001...

Hi Dagon, thx for yr reply. yes you were right the indexes of session_seller_tbl were not actual. Now the response time of sessions_tbl similary to sessions_vw, but I forgot an important think. The whole sessions has to be ordered by starttime cause I allways need only the latest 3000 because...

Hi guys I have a performance problem and no idea how to handle it ;( I will illustrate it by an example: there are 4 tables : one contains the actual session data [session_tbl (2 millions ds, may exeeds to 24 millions)], some sessions are linked with some sellers [seller_tbl (1000 ds)] the...