Re: [PERFORM] Heavy contgnous load

2011-11-13 Thread kzsolt
I try to found out more information. Looks like the COUNT(*) is not the strongest part of pgr therfore I do a worakround. After this I have the folwing result: Below the 1Mrec the row insert time is ~23msec. Above the 7Mrec the insert time is ~180msec. I belive I use the fastest index type (def

Re: [PERFORM] Heavy contgnous load

2011-11-11 Thread kzsolt
Looks like I found more magic. My table is: each record near 1kbyte, contain dozen col some text some numeric, some of the numeric columns are indexed. The database located at ramdisk (tmpfs) ((I hope)). The table is contignously filled with rows. If the table has less than 4Mrec then looks like

Re: [PERFORM] Heavy contgnous load

2011-10-23 Thread kzsolt
So guys, lot of thank you for all of the explanation and ideas! Jeff Janes wrote: > What happens if the database has a hiccup and can't accept records for > a few seconds or minutes? Craig Ringer wrote: > If you really need absolutely maximum insert performance, you should > just use a flat fil

Re: [PERFORM] Heavy contgnous load

2011-10-19 Thread kzsolt
Kevin Grittner wrote: > Rollback activity? What rollback activity? When you're doing what? > What is the exact message? I mean here some kind of option to save reources. For example mysql has table (storage) type where no transaction support (rollback) in. This make the all server faster and us

Re: [PERFORM] Heavy contgnous load

2011-10-19 Thread kzsolt
"try to have as few indexes as possible and do your inserts inside transactions in batches, rather than one-by-one as individual statements. " That is the main problem. I use now few index as possible. Unfortunately the one-by-one INSERT is nature of or system. To join (batch) inserts is require sp

[PERFORM] Heavy contgnous load

2011-10-18 Thread kzsolt
Dear Anybody! I use pgr to store records. But the characterisitc of the record traffic are special. For example 50 of them arrived in one sec contignously trough weeks and aligned interally trough tables. To absorb this traffic I put the pgr database to ramdisk (fast as possible). But after more