Обсуждение: Multi-tenant cluster

I've done some searching using Google and found a few papers and
articles on multi-tenant databases.  I still have to read through many
of them but there didn't appear to be much reference to using PostgreSQL
in the ones I found.

Can anyone suggest sources of information on this topic -- multi-tenant
databases?


Thanks,
Rod
--

On Wed, Dec 24, 2008 at 8:17 AM, Roderick A. Anderson <raanders@acm.org> wrote:
> I've done some searching using Google and found a few papers and articles on
> multi-tenant databases.  I still have to read through many of them but there
> didn't appear to be much reference to using PostgreSQL in the ones I found.
>
> Can anyone suggest sources of information on this topic -- multi-tenant
> databases?

PostgreSQL can do this quite well.  There are a few ways of
approaching this.  The most separation can be achieved by setting up a
db per user and configuring pg_hba.conf for sameuser so that the user
connects to the database named for them only.

Scott Marlowe wrote:
> On Wed, Dec 24, 2008 at 8:17 AM, Roderick A. Anderson <raanders@acm.org> wrote:
>> I've done some searching using Google and found a few papers and articles on
>> multi-tenant databases.  I still have to read through many of them but there
>> didn't appear to be much reference to using PostgreSQL in the ones I found.
>>
>> Can anyone suggest sources of information on this topic -- multi-tenant
>> databases?
>
> PostgreSQL can do this quite well.  There are a few ways of
> approaching this.  The most separation can be achieved by setting up a
> db per user and configuring pg_hba.conf for sameuser so that the user
> connects to the database named for them only.

Thanks Scott.  Sorry to take so longing getting back.

This is a little too fine of a granularity for what I'm trying to do.

I have (will have) some databases in a cluster that represent different
tenants (company's, organizations).  In each database there will be
several schemas that hold the data for an application that is typically
done as a single database, in the public schema, in a cluster.


The thorniest issue I've come across is user(s) names for each
application in each database.  Plus a dba account per database.

Currently I'm using db1_dba, db1_dba, db1_rt_user, etc.  I understand I
can do something like rt_user@db1, dba@db2, ... but there are or could
be some problems with doing it this way.

So back to the books for me.


Again thanks,
Rod
--