Configuration - PostgreSQL 9 Administration - page 61

F user simon connects to a database other than saas, then value2 will apply
F user simon connects to database saas, then value3 will apply
PostgreSQL implements this in exactly the same way as if the user had manually issued
the equivalent SET statements immediately after connecting.
Basic server configuration checklist
PostgreSQL arrives configured for use on a shared system, though many people want to run
dedicated database systems. The PostgreSQL project wishes to ensure that PostgreSQL will
play nicely with other server software, and should not assume it has access to the full server
resources. If you, as the system administrator, know that there is no other important server
software running on this system, then you can crank up the values much higher.
Getting ready
Before we start, we need to know two sets of information:
First, we need to know the size of the physical RAM that will be dedicated to PostgreSQL.
Second, we need to know something about the types of applications for which PostgreSQL
will be used.
How to do it...
If your database is larger than 32MB, then you'll probably benefit from increasing
shared_buffers . You can increase this to much larger values, though remember
that on Linux systems this memory can be swapped out if not in use, so it's better to be
conservative. A new value can be set in your postgresql.conf and incremented slowly
to ensure you get benefits from each change.
If you increase shared_buffers , and you're running on a non-Windows server, you will
almost certainly need to increase the value of the OS parameter SHMMAX (and on some
platforms others as well).
On Linux/Mac OS/FreeBSD, you will need to either edit the /etc/sysctl.conf file or use
sysctl -w with the following values:
F Linux: kernel.shmmax=value
F Mac OS: kern.sysv.shmmax=value
F FreeBSD kern.ipc.shmmax=value
http://www.postgresql.org/docs/8.4/static/kernel-resources.
html#SYSVIPC