| List: | General Discussion | « Previous MessageNext Message » | |
| From: | Perrin Harkins | Date: | July 23 2007 6:47pm |
| Subject: | Re: performance of extended insert vs. load data | ||
| View as plain text | |||
On 7/23/07, mos <mos99@stripped> wrote: > Load data will of course be much faster. However to obtain the maximum > speed you need to load the data to an empty table, because then MySQL will > load the data without updating the index for every row that's added, and > will instead rebuild the index only after all data has been loaded, which > of course is much faster. He's using InnoDB tables, so it doesn't make much difference. MyISAM tables have a fast bulk index rebuild, but InnoDB do not. > The other alternative is to remove all indexes > prior to using load data and build the index manually when the data has > been loaded. The easiest way to do that is with ALTER TABLE DISABLE KEYS and ENABLE KEYS. But again, not much help with InnoDB tables. - Perrin
| Thread | ||
|---|---|---|
| • performance of extended insert vs. load data | Sid Lane | 23 Jul |
| • Re: performance of extended insert vs. load data | Perrin Harkins | 23 Jul |
| • Re: performance of extended insert vs. load data | mos | 23 Jul |
| • Re: performance of extended insert vs. load data | Perrin Harkins | 23 Jul |
| • ordering dates | Ross Hulford | 12 Sep |
| • Re: ordering dates | Michael Dykman | 12 Sep |
| • Re: ordering dates | Philip Hallstrom | 12 Sep |
| • RE: ordering dates | Jerry Schwartz | 12 Sep |
| • Re: performance of extended insert vs. load data | B. Keith Murphy | 23 Jul |
| • Re: performance of extended insert vs. load data | Mogens Melander | 24 Jul |
| • Re: performance of extended insert vs. load data | Ravi Prasad | 24 Jul |
