Re: Changing the concept of a DATABASE

On May24, 2012, at 11:39 , Susanne Ebrecht wrote:
> There are lots of offices / departments creating maps. Topography maps,
> pipeline maps, nature conservancy (e.g. where are the nests from endangered
> birds?), mineral resources, wire maps, street maps, bicycle / jogging maps,
> tourists maps, tree maps, cadastral land register, and so on.
>
> All this departments have their own databases for their own maps.
> They only map their own stuff.
>
> Towns / states / regions have a department where all these maps get collected.

The question is, how do they get collected? If they use some home-grown replication,
they might just as well collect them into schemas instead of databases. The same is
possible with slony, I think. And if they use WAL-based replication, they have no
choice but to collect them in different clusters, so cross-database queries within
a cluster wouldn't help.

I think that you're right that reporting would one of the main use-cases for cross-
database queries. But reporting is also, I think, one of the main uses-cases for
WAL-based replication. So having cross-database queries with don't allow queries
across multiple replicas will leave quite a few people out in the cold.

best regards,
Florian Pflug