Does postgres include this feature? Can someone point me toward the docs?
This is regarding the old clustering thread. If the on-line replication is
implemented, then it could be used for a form of clustering, because the
SELECTs could be distributed across multiple servers (one on each server).
Obviously, this would require the complete database to be on each server,
but until more advanced clustering methods are implemented, I don't see
another sensible way of doing clustering for situations where one server
just cannot feasibly cut it. Besides, insuring completeness of the returned
results when doing a multi-table join across multiple servers is a mind
numbingly hard problem...
I am also guessing that record locking would not work properly if multiple
computers were accessing the same physical database file that is mounted on
a remote file system. Is this the case? Not that I am even thinking about
trying to access data files via NFS or DFS... ;-)
Regards.
Gordan