Paul Rauch wrote: > Is there a way to reduce execution time by counting objects? > > it's even splitted in two queries now, with only 2474487 entries ;) > > in django I use "model.objects.count()" > > SELECT COUNT(*) FROM "rainbowtables_hashes":34.552 > > SELECT COUNT(*) FROM "rainbowtables_hashes":31.024 >
I guess you are using postgresql. This is a known "issue" with that database - count(*) on an entire table (or large subsets) is slow (and takes time proportional to the rows found). See a discussion of the reasons here: http://archives.postgresql.org/pgsql-hackers/2005-01/msg00247.php Basically the developers are disinclined to "fix" this, since it would make each index bigger, slowing down other operations. Some people stated that they were going to work on creating this as an alternative index type, but AFAIK it hasn't happened yet. A workaround could be to store a copy of the id in a one-to-one related table (id is Primary Key AND Foreign Key to rainbowtables_hashes. Then you can just count the contents of that table (or use it in a join if you need more criteria). I am not sure what the easiest way of getting Django to do this for you though. Hmm, i just realized a generic way of making this work in postgresql (if you are happy to do your own ddl). Basically use an updatable view, which maintains the additional table behind the scenes. Will write that up in more detail later. /Nis --~--~---------~--~----~------------~-------~--~----~ You received this message because you are subscribed to the Google Groups "Django users" group. To post to this group, send email to django-users@googlegroups.com To unsubscribe from this group, send email to [EMAIL PROTECTED] For more options, visit this group at http://groups.google.com/group/django-users?hl=en -~----------~----~----~----~------~----~------~--~---
