Re: Rép. : Re: start/stop a database

On Fri, 9 Jan 2004, John Sidney-Woollett wrote:

> Hi
>
> My understanding of the differences between Oracle and Postgres is
> slightly different:
>
> Oracle        vs      Postgres
> ------------------------------
> database       -      cluster (see note i)
> instance       -      database
> user/schema    -      schema  (see note ii)
> user           -      user

That looks about right.

> Some other things that you might miss from Oracle in postgres are:

SNIP

> 7) No fine grained control on where your objects are stored (tablespaces,
> data files), but apparantly coming soon

The closest thing Postgresql has is locations, which allows you to put a
whole database instance in a different location than the rest of the
database.  Crude, but useful for certain things.

Also, if you're willing to shutdown the database and move and then
softlink the indexes or tables you can get something like fine grained
control, but reindexing or dropping and recreating an object will result in
the index moving back to the original location.  It's a hack and a kludge,
but one that can sometimes be a good enough solution.

> There are probably some others, that I can't think of right now. Most
> things that you will miss, you can probably work around without too much
> difficulty by adopting a slightly different mindset.

The other issues are often more subtle, like the fact that there is no
rollback segment, so there is no limit to a maximum transaction other than
disk space.  Most of these differences are only a problem because "the
oracle way" is different, not necessarily better or worse than, "the
postgresql way".

> Apologies if anyone feels that I keep harping on about Oracle, but it is
> my main DB/SQL frame of reference for the moment.

No offense taken.  The Postgresql developers, rightly so, believe Oracle,
DB2, and that class of database as their major competition, so it's not at
all uncommon to look at how oracle or db2 does things when deciding on
how to do things.