On Fri, Sep 20, 2013 at 4:38 AM, Julian <tempura@internode.on.net> wrote:
> However, I tend to go with partitions when required to be generated on
> demand dynamically and automatically (which probably isn't the case
> here). SCHEMAs have other uses, provide a level of security (GRANT) and
> useful in design when partitioning off blocks of related datasets
> completely.
I would do a partition in this case for the same reason: schemas are
much more logical divisions of data due to different grantings, search
paths, users, and so on. Partition is more a data-level split so it
does make sense when you want the data to stay in a single point but
for performance reason split it across multiple tables.
Luca
