On 23/10/09 18:25, Anthony wrote: > On Fri, Oct 23, 2009 at 1:11 PM, Tom Hughes<[email protected]> wrote: >> In the real world we don't all have infinite CPU time and I/O bandwidth > > Hence my point about using a modern database, where you can create > things like materialized views to make accessing the database > efficient without making the core schema itself a mess.
I understand the concept of a materialized view, but I've never actually used one myself. There is of course a significant cost in terms of disk space if you create materialized views with a large data set like ours. One other obvious question is how materialized views interact with updates to the underlying tables? Do they have to be recomputed from scratch every time an underlying record is changed? Only it seems that solving the generalised problem of working out how to update a view given a specific change in the underlying data would be a hard one to solve. Tom -- Tom Hughes ([email protected]) http://www.compton.nu/ _______________________________________________ Geocoding mailing list [email protected] http://lists.openstreetmap.org/listinfo/geocoding
