There is a lot of going back and forth about shared memory buffers and postgres
working "out of the box."
This is sort of a discussion I'd like to see. I've been using Postgres for a
long time, several years and in a few jobs, it is an excellent product, with a
depth of configurability that most new users do not know. For these users,
this lack of knowledge sometimes prevents them from using Postgres because they
do not perceve that it takes a little configuration adjustment to work right
for an application.
Packages like Oracle and MS SQL are a nightmere to set up, but they do REQUIRE
that you are familiar enough with some settings. More to the point, they
REQUIRE that you answer some questions about how you will be using it during
setup.
I'm not suggesting that Postgres adopt the user hostile stance or Oracle, but
the "out of box experience" of simply working, may not, from a user interface
perspective, be the right one.