[Pgpool-general] Understanding the connection pooling

[lst_hoe01 at kwsoft.de]

Zitat von Tatsuo Ishii <t-ishii at sra.co.jp>:

>> we look for some way to limit the number of active PostgreSQL server 
>> processes
>> with clients doing mostly idle persistant connections to the DB. On the
>> PostgreSQl mailing-list someone pointed me towards pgpool but from 
>> what i see
>> (http://lists.pgfoundry.org/pipermail/pgpool-general/2005-February/000036.html)
>> this isn`t the right tool to get something like "shared server" with 
>> Oracle or
>> have i missed something?
>> Can pgpool reduce the number of server processes per client?
>
> What is Oracles's "shared server"?

Sorry for being imprecise. Oracle shared server (MTS) is a dispatcher in front
of the server processes which hold endpoint of the client connections and
multiplex the client queries to a limited (lower) number of backend worker
processes. The goal is to support a huge number of concurrent client
connections without needing the overhead of one server process per client
connection.
Our scenario is a CTI application using one persistant ODBC connection to the
server per addressbook with four of them available for every user...
So the "query load" per connection is very low, but the number of 
connections to
the server is quite high.

Regards

Andreas