senor <frio_cervesa@hotmail.com> writes:
> Is the limitation simply the state of development to date or is there
> something about dumping the schemas that conflicts with paralleling?
At minimum, it'd take a complete redesign of pg_dump's output format,
and I'm not even very sure what such a redesign would look like. All
the schema information goes into a single file that has to be written
serially. Trying to make it be one file per table definition wouldn't
really fix much: somewhere there has to be a "table of contents", plus
where are you going to put the dependency info that shows what ordering
is required for restore?
> The --link option to pg_upgrade would be so much more useful if it
> weren't still bound to serially dumping the schemas of half a million
> tables.
To be perfectly blunt, if you've got a database with half a million
tables, You're Doing It Wrong.
regards, tom lane